#qi-hardware IRC log for Sunday, 2012-01-29

lindi-_GNUtoo: ublox antaris 400:00
GNUtoois it an ARM soc?00:00
lindi-_yes, I've run code inside it00:00
GNUtooI tought you patched 1 byte of it00:00
lindi-_yes that was what I published00:00
GNUtooto enable complete traces00:00
lindi-_I can't actually distribute patches to non-free firmware :/00:00
lindi-_but I hope a single byte can't be infridgement00:01
lindi-_anyways, ublox offered an SDK years ago for developing third-party software to run on the GPS chip00:01
GNUtooimagine a freerunner with osmocombb on top of nuttx-bb + a free GPS firmware....that would be awesome00:01
kristianpaulGNUtoo: if you reverse eng the firmware run inside the bcm4751 soc, that will be very awesome !00:02
kristianpaulhere are some links http://en.qi-hardware.com/wiki/GPS_Free_Stack/Web_Links00:02
kristianpaulvery messy tought..00:02
wolfspraulGNUtoo: can you motivate some people to hack on Milkymist One and join kristianpaul's free GPS baseband project? that'd be awesome too! :-)00:02
lindi-_does milkymist one have any external antenna?00:03
lindi-_internal gps antenna won't make RTK possible00:03
kristianpaulbut if you have more information about this bcm i can try help/point sopmething00:03
kristianpaullindi-_: mine does :)00:03
kristianpaulGNUtoo: you want to calculate fix on the CPU? get a port of osgps for your OM, and try to hookup a gps receiver00:04
kristianpaulgps receivr IF, like those from SigE or Maxim00:04
kristianpaulbut you alredy said no to hack the firmware/arm cortex part of this bcm4751, too bad :(00:05
GNUtooI would prefer to hack the GPS firmware of the freerunner....00:06
kristianpaulbut lindi-_ already did it no?00:06
kristianpaulcalculate fix using rtklib00:06
GNUtooah ok00:06
kristianpaulfrom navigation data provided from the ublox chip00:06
kristianpaulhack the bcm4751 is more interesting because you still chance to implement basebadn00:07
kristianpauljsut as wolfspraul said, thats what we are working on00:07
kristianpaulmust go, need go grocery store00:08
GNUtooI must go to sleep btw00:08
kristianpaulsure ,)00:08
kristianpaulMAX2769 -> OM00:09
kristianpaulthe OM calculate in real time a fix :)00:09
kristianpaulosgps does it00:09
kristianpaulnot hard !00:09
kristianpaulwell missing real time part, but still way to go most in software00:10
kristianpaulbaseband moves just to software00:10
kristianpaulnot HDL  and Software like ours currently00:10
kristianpauldunno who will go faster00:10
kristianpaulok by GNUtoo !!00:10
rohkristianpaul: my guess is that we will combine stuff more and more to have 'multiple purpose cpus' which have the abilities of a gpu, a cpu, and a vector cpu (e.g. dsp stuff)01:37
rohthe arm vector calculation stuff could make sense for such stuff... dunno exactly what kind of math is needed01:37
kristianpaulmath is fft in parallel for signal acquisition and tracking01:41
kristianpaullater fix i resolve a matrix dont remenber of wich dimensions01:42
kristianpaultalking about a full software aprouch01:42
kristianpaulwell ignore full, we still using IF :)01:44
wpwrakthat new feature of qi-bot of sucking the life force out of any newcomers joining the channel seems to work :)05:45
xakhI lost the little rubber light thing05:55
xakhis there a way to buy a new one?05:55
xakhoh wow, sorry about how stupid that sounded.05:56
xakhOn my nanonote, my girlfriend helped me take it apart, because for the past two years, there's been a piece of graphite lunking around in there.05:57
xakhit hadn't bugged me much05:57
xakhbut suddenly, the graphite moved up and shorted out the board.05:57
xakhso, she helped me open up the case, and the thing fell out, eventually (it was stuck UP there), and we got it out05:58
xakhwhen it was put back together, the little rubber piece of plastic that covers the LED05:59
xakhwas gone, and we can't find i.05:59
xakhI know it functions fine without it, but at the same time I worry about dust, fuzz, etc getting in there.05:59
xakhso is there a way for me to buy a new rubber piece for the light?06:00
xakhalso, is there a way for me to buy qi-hardware stickers?06:00
xakhbecause you guys have the neatest logos.06:00
xakhlike, I'm specifically looking for the Ben symbol in a circle, like what it has when it boots.06:01
xakhbecause that looks killer.06:01
wpwrakhm, you'll have to wait for wolfgang for an official replacement and stickers06:02
wpwraka quick fix for the led rubber would be with transparent silicone06:02
wpwrakjust put a drop there, then shape it with your finger and swipe off the excess material06:03
xakh....where would I get that?06:03
wpwraksilicone can be shaped easily if you wet the finger (or other tool) with isopropyl alcohol06:03
wpwrakhmm. any place that has plumbing supplies should have silicone06:04
xakhah alright06:04
wpwrakhardware store, etc.06:04
xakhwell, still, I always like to support you guys, and buying one from you guys would be preferable06:05
wpwrakor the big supermarkets where you can get all sort of DIY thing06:05
xakhalso, if I get one of those, there's a damn good chance I'm gonna gum up my nn06:05
wpwraksilicone is easily removed :)06:05
wpwrakit's a pretty neat material for little DIY jobs06:06
wpwrakwell, and you can seal your bathroom with it, etc. ;-)06:06
xakhalso, is there a place in the US that sells the atben stuff? because I feel like the shipping from tuxbrain is going to be huge06:06
xakhwell, like I used to use Sugru with stuff, is it like that?06:06
wpwraki don't know. the only ones i've heard of are tuxbrain and pulster06:06
wpwrakyes, like sugru but more liquid06:07
xakhactually, until today, Sugru was what made up the little feet on my nanonote06:07
wpwrakhehe ;-)06:07
xakhbut it was in the way of a screw, so they had to go06:07
wpwrakyeah, a tinkerer's device should not hide the screwd under rubber feet. instead, it should show them with pride ! :)06:08
wpwraki think sugru is actually based on silicone. they just made it a bit tougher.06:09
xakhyeah, I think so too06:09
xakhnext time I get some more sugru though I'm making some new feet06:10
wpwrakyou can probably also use silicone for that. cheaper :)06:11
wpwraksilicone comes in colors. transparent, white, and black should be easy to get. but maybe you have a much larger choice.06:12
xakhtrue, but silicone doesn't dry into as frictional a surface, like my nanonote wouldn't slide down a slanted surface, at all06:14
xakhI had a plastic repair kit somewhere06:15
xakhthat I think was silicone based06:15
xakhand it was really smooth06:15
wpwraksilicone has a rubberish texture06:17
wpwrakso i think it should be fine for feet. but you can just try whether it works. it's not expensive.06:18
wpwrakthere's also a stronger variant, polyurethane. that's a bit trickier to handle, though. and you won't get it off things as easily :)06:19
xakhalso though, still wanna buy stickers06:20
xakhand maybe an atben thing, but that's for tuxbrain06:20
wpwrak(stickers) yeah, they probably don't have these at the DIY store :)06:21
wpwrakah, maybe wolfgang can also sell atben/atusb. i think he once thought about it. but i'm not sure if anything came out of it.06:21
xakhthat's no big deal, I gotta figure out how to afford one, but other than that I'm fine06:22
xakhbut either way, should I just shoot a message to the mailing list?06:30
wpwrakyeah, why not. or hang our here until wolfgang shows up.06:35
xakhalrighty, sounds good!06:35
wpwraki think he's still in germany. it's 07:35 there. may take a few more hours :)06:36
xakhahh alright06:37
xakhman, that guy's a globetrotter.06:37
xakhwhere does he actually live06:37
wpwrakin beijing06:38
xakhahh alright06:39
xakhthe name wolfgang threw me06:39
xakhor is it a handle?06:39
wpwrakno, he's german. beihin is just the place he chose to live at for now06:41
xakhhaha, that's an awesome bot.06:42
xakhdang it, he had a cool website, I was just on06:43
xakhbut I forgot it.06:44
xakhyou have any idea where to find it?06:48
xakhoh wait no06:49
xakhthat was xiangfu.06:49
xakhdunno why I thought it was wolfgang, I guess I always just lump those guys together, since I always talked to them together or something.06:50
wolfspraulxakh: hi good morning07:37
wolfspraulI was planning to make exactly those round Ben stickers, but as of today I don't have them yet07:38
xakhhi wolf!07:39
xakhthose sound amazing.07:39
wolfsprauland I don't have these tiny led covers either, though maybe I can take one from a damaged unit07:39
wolfspraulthat's Jon's design, he will like you liking them :-)07:39
xakhI love those.07:40
wolfsprauldoes anyone know more about http://www.elinux.org/Embedded_Open_Modular_Architecture07:40
wolfspraulroh was asking in #milkymist, but I thought I ask here too07:40
xakhI know a bit about compactFlash cards, but that's it.07:41
xakhI know that's not really helpful07:41
xakhbut it's what I know07:41
xakhso why are you talking about the EOM?07:46
wolfsprauljust looked at it because roh was asking07:46
xakhah ok07:47
wolfspraulxakh: where are you located?07:56
wolfspraulI will definitely remember you when I have the round Ben stickers, although it may take a while07:57
xakhI'm in the US08:18
xakhsorry for the late response08:18
xakhthat's fine, I'm happy to get them any time08:18
nickoeHi wpwrak, how do you make a footprint, with pads that is not only rectangular or with rounded corners? I get an error if I try to make two pads overlap.08:44
nickoeI am talking about fped.08:44
wolfspraulgood morning11:19
wolfspraulmethril: how is life in Brazil? (you are there, right?)11:19
methrilwolfspraul: i'm not in brazil right now11:21
methrilwolfspraul: this why i've been missing 11:21
methrilwolfspraul: i'm back to Europe (Ireland)11:21
wolfspraulahh :-)11:23
wolfspraulbut... returning to Brazil soon?11:23
wolfspraulI hope you follow Qi and Milkymist still :-)11:24
wolfspraulwe are on a collective learning experience I feel11:24
methrili moved to Ireland, i'm not going to return to Brazil for living11:24
methrili follow Qi and Milkymist11:25
wolfspraulso you've been missing Ireland?11:25
wolfspraulhow did it go in Brazil?11:25
wolfspraulI didn't know you moved back, haven't seen any blog posts from you in a while11:25
methrilyes i know, i've been lazy about updating the blog11:26
methrili received a good offer here and i accepted11:26
methrili'm planning to do a post, but i would like to have time to work on something interesting before putting something in the blog11:27
wolfspraulwhat was your experience in Brazil overall?11:33
wolfspraulhow about foss there?11:33
DocScrutinizergot another goodie for you, hope the mail will come thru (related to crowbar discussion yesterday)11:38
methrilthe FOSS dev is not really bad. I think that a lot of companies use FOSS developments, but it didn't differ so much from Euroope11:39
methrilIMHO the companies use FOSS for their own benefit, and they only share the code when they are forced to do it11:40
methrilBTW i enjoy living in Brazil. Was a nice experience11:40
DocScrutinizernow moving to Ireland in winter... :-/11:41
DocScrutinizerhope you are a snow lover11:41
DocScrutinizerI hate it11:42
methrilDocScrutinizer: no snow in Ireland 11:42
DocScrutinizersure, but still nasty weather11:42
methrilonly rain... rain..... and more rain11:42
Action: methril nods 11:42
DocScrutinizerlike in whole Europe this year11:42
DocScrutinizerwinter got cancelled this year, we got a 6 months of november11:43
methrili prefer Spain weather, but... my new work was in Ireland ;)11:43
DocScrutinizerglobal warming \o/11:43
DocScrutinizeryeah, "winter" at costa de la luz is really really bearable :-D11:44
methrilbut Europe is small compared to Brazil11:45
methriland the flights are cheaper11:45
methrilyes, the countries in Europe are "small"11:47
DocScrutinizersigma-Europe not so much11:47
wolfspraulso what is your new work about? foss-friendly company?11:48
DocScrutinizerI think it's about 10.000km from upper north down to gibraltar or greece11:48
methrilnot so much foss-friendly11:48
methrilthis is the pitty11:48
wolfspraulI am quite pleased how the NanoNote and Milkymist are chugging along, though we still need some luck in reaching critical mass and making a real hit product11:48
methrilis a Processing Image company11:49
wolfspraulif you have any input/feedback, please holler, we need to hear it11:49
DocScrutinizerif you're counting Melilla of Spain to Europe then it's probably even more ;-D11:49
methrilDocScrutinizer: but they are different countries11:49
methrilBrazil is only one11:49
DocScrutinizerGreece is quite a bit different nowadays ;-P11:50
methrilanyway, Europe has other things that Brazil doesn't have11:50
methrili didn't know Greece yet11:50
Action: DocScrutinizer neither11:50
DocScrutinizerit's with the 50% European countries I haven't visited yet11:51
DocScrutinizerthat might cure soon without me taking any action ;-P11:52
methrilyes, there are a lot of counties i haven' visited also11:52
methrilwhat kind of action?11:52
DocScrutinizerit's going to simply be no member of EU soon11:52
DocScrutinizernot really11:53
methrili also think Spain is in a bad situation11:53
DocScrutinizeryeah, internally sure. But not that much regarding their credibility11:54
larscireland too11:54
methrilyes, ireland too11:54
DocScrutinizerireland already recovers I rhink11:54
methrili also think ireland recovers11:54
DocScrutinizeronce we nuke Standar&Poors with a cruise missile, our problems here are mostly solved ;-P11:55
DocScrutinizerbut wait, that's the American Way to solve problems11:56
methrilGermany is ok, isn't it?11:56
DocScrutinizermostly, yeah11:56
larscgermany is the big winner of the whole crisis11:57
DocScrutinizerpaying more than we have for saving other countries11:57
larscgerman government bonds have negative interest11:57
larsci.e. banks pay money so can lend money to the german state11:57
DocScrutinizerlarsc: this will change in no time, once Greece needs triple the money they asked for now11:58
DocScrutinizerand Italy joins in11:58
larscdon't think so11:59
methrilif Greece goes out of European Union, how are they going to force to pay? They don't have political/economic way of pressure12:00
DocScrutinizerGermany can't sustain their brilliant export rates when other countries in EU don't want to buy expensive stuff  anymore from us. Same time we can't refuse to help stop anarchy from state bankrupt in Greece and other countries, so we HAVE to pay LOTS12:01
C-Keenthis is not about money but control12:02
DocScrutinizeror lack thereof, in *all* governments12:02
DocScrutinizerit's already the big biz that rules, and did so for some decades now12:03
methrilbig OT in qi-hw \o/12:04
DocScrutinizer"Money is a gas"12:04
methrilmoney is the gas12:04
methrilif you don't have money, you have no global pressence12:04
DocScrutinizer(OT) well, according to my own IRC rules that are loosely inspired by freenode's rules, there's no such thing like OT until somebody shouts "OT!!"12:05
DocScrutinizerthe very moment somebody does, I'll enforce going on topic again on any chan I got +o12:06
methrilDocScrutinizer: you could jump some rules today ,)12:08
Action: DocScrutinizer considers adopting the practice to insert a ====================== 8X (cut here) ====================== line when doing so, for the convenience of users reading thru their backscroll12:09
DocScrutinizermethril: I haven't heard a serious complaint yet12:09
DocScrutinizerneither osme user trying to get thru with a on-topic question12:10
DocScrutinizerwhich would trigger same procedure12:10
methrilwell, what are you working on?12:10
DocScrutinizerplease rephrase the question12:11
DocScrutinizerI'm always working on a lot of things :-D12:11
methrilok, i know12:11
methrili was you doing a lot of hw hacks12:12
DocScrutinizerjust posted a suggestion supplement to the crowbar idea of yesterday12:12
methrilI have been out for 2/3 months12:13
methrilthis is why i'm asking12:13
DocScrutinizerI bet wolfspraul or wpwrak can share a pipermail URL to the relevant thread on mikymist ML12:14
DocScrutinizer(I'm not even sure my reply to wpwrak 's post made it to the ML. Probably "pending approval")12:14
DocScrutinizernot a representative topic regarding what happened last 3 months here though12:15
DocScrutinizerI'm not very active here, just helping a bit every other month12:15
DocScrutinizerusually on request12:16
DocScrutinizergot a time demanding daywork since 2 months now - building STE's LTE modems12:17
DocScrutinizeror rather: getting used to such crap like winXP, ClearCase, similar abominations12:18
DocScrutinizeractually I started "productive work" last Monday12:18
DocScrutinizeralready contributed an awesome ~30 lines of code, to fix an prio:low internal ticket12:19
DocScrutinizerthanks to 4 days studying code and mainly c&p "editing" they were almost initially bugfree \o/12:20
DocScrutinizeractual implementation took only 2 hours, after those 4 days of code study12:21
DocScrutinizerincl fixing the one typo (missing ";") and test12:21
methrilDocScrutinizer: you are working still on GTA0X devices?12:24
DocScrutinizerwell, I try to help with goldelico's GTA04, but again there's much reluctance to consider my concerns and suggestions12:25
DocScrutinizerand poor communication/workflow-integration (I got no access to mist recent schematics for example)12:26
methrili though GTA04 was "open" in dev12:26
methrili didn't follow the ml for months (more than 3)12:26
methriland i read something about it the last week12:26
wolfspraulmethril: is your Milkymist One in Ireland now?12:27
wolfspraulor went MIA somewhere?12:27
DocScrutinizeryeah sure, the Eagle professional project files are probably available - I'm not going to buy Eagle-Pro to do some voluntary edits or just look at schematics12:27
methrilwolfspraul: is in Ireland12:27
methrilthis is why KiCAD could help on this12:28
wolfspraulyes we plan some heavy KiCad action for M1 soon, and Joerg's input will be very much appreciated12:28
DocScrutinizerthe "fun" part is the constant deja-vu I got with GTA04 - reproducing all the mistakes and misconceptions we did on GTA02 before12:28
methrilwolfspraul: any Ya plans in the near future?12:28
wolfspraulno, not near future12:28
wolfspraulthat would require Ingenic to become more active which so far has not been happening12:29
wolfspraulso I power forward with Milkymist One instead, eventually these things should merge anyway12:29
methrilthe pity is not hearing people that knows about those mistakes12:29
wolfspraulthey don't want to hear, need to be so blunt12:29
DocScrutinizerrecently they noticed their modem doesn't power down on system shutdown - OMG12:30
wolfspraultheir speed is let's say 10,000 times too slow12:30
DocScrutinizeraudio section messed up it even beats GTA02 in that regard12:30
methrilit looks every uP is trying to put Android on int12:30
wolfspraulit's like 3 snails trying to jump on a 250km/h high-speed train12:30
wolfspraulfun to watch though how they stand on the bridge planning :-)12:30
DocScrutinizerwell, Nikolaus managed to produce a rev3 PCB that already works better than any of OM's GTA01 ever did ;-P12:32
wolfspraulDocScrutinizer: they cannot process your input because the ultimate realization would be that their entire trajectory is mission impossible12:32
Action: DocScrutinizer tends to agree to some degree12:32
wolfspraulfine then, he will have another 1000 things ahead of him, if each takes him let's say 2 months, that's 2000 months that's ...12:32
wolfspraul166 years12:33
DocScrutinizerthe nice part is this isn't planned as any commercial success (almost wrote "sucksess" ;-D)12:34
methrilthis is why are so expensive12:34
methrili prefer to spend my money in other "toys"12:34
DocScrutinizerOM did PV's of higer volume than what doldelico plans for first sales batch12:34
methrilthat could work12:34
wolfspraulmethril: the issues go far far deeper12:35
wolfspraulso much that it's just silly to even discuss12:35
wolfspraulJoerg knows what I mean12:35
wolfspraulwe cannot break this down to a 10 minute irc chat12:35
DocScrutinizerstill I like to help12:35
wolfspraulJoerg has what... 30 years industry experience?12:35
methrili understand12:35
DocScrutinizerumm, only passive12:35
wolfspraulI have 5 (in hardware). I know nothing, but at least I get up every morning beating myself into shape to learn more that day, why the damn things are not working out.12:36
DocScrutinizermy industry experience always been "on your side of the monitor"12:36
methrili don't think you have only passive, i know you are a really good engineer12:36
wolfspraulyes but hardware is very subtle on the cost and time dimension12:36
methrili saw your fixes and ideas on e-mails12:37
DocScrutinizersure, but that'S not related to "industry"12:37
wolfspraulso just watching for a number of years, with a fresh mind, will already be good12:37
methrili lear from both of you :)12:37
wolfspraulI'm a newbie12:37
wolfspraulI remember sitting still for hours though watching the FreeRunner assembly line12:37
methrilif you are a newbie.... i have no words for me lol12:37
wolfspraulthat was so painful12:37
DocScrutinizerI'm closely watching what my EE and chip brewer colleagues did last 40 years12:38
wolfspraulI see all these things going on, and I can't believe what I see.12:38
methrilthe industry changed a lot...12:38
methrili think it evolves really fast12:38
DocScrutinizerthat's why I'm good (maybe) - I never got limited to one narrow topic12:38
wolfspraulbut it was reality, so I kept digging deeper12:38
wolfspraulanyway, I do wish gta04 luck12:38
methrili also wish luck12:39
methrilbtw, who is getting good steps is Harald with the Osmocon project12:39
wolfspraulmaybe we can work together one day? don't know. probably not, since their thinking seems really stuck.12:39
wolfspraulthey need a miracle! :-)12:39
wolfsprauloh sure, can't even compare12:39
DocScrutinizerwell, realistically comparing it to e.g. pandora, goldelico doesn't perform bad at all12:39
wolfspraulthose are building blocks for the future12:39
wolfspraulha ha12:39
wolfspraulI agree12:39
wolfspraulbut if that's all that open hardware could ever be, i would happily use my iphone and be in another industry12:40
DocScrutinizerpreaching to the choir12:40
methrilthis is true, the Open Hardware industry didn't reach the mass12:40
wolfspraulit will12:40
wolfspraulin my opinion that's a matter of understanding the process better (design and manufacturing and qa and...) and getting the economics right12:41
methrilthe last part is what goes some project to a no way end... so it's important12:42
wolfspraulmany projects seem to be caught in a hero complex12:42
wolfspraulthey want to be heroes, hero designers, etc. the genius who created xyz12:42
CcSsNETsadly open hardware is still highly imperfect12:42
CcSsNETsuch as most devices using SD ports12:43
wolfspraulbut hardware in my opinion is about manufacturing economically, serving customers with actual physical goods12:43
CcSsNETdrm infested storage.... wth...12:43
wolfspraulthat won't win you any hero award12:43
CcSsNETwhats up with your obsesion of fame?12:43
methrilit depends, for me the technical guys like lekernel, DocScrutinizer, larsc, you ... are almost heroes ;)12:44
DocScrutinizerOT: you think it's a good thing to be early adopter and get <lastname>@<anymail>.<fld> ? Believe me it's not, all the idiots misspell their own email addr <firstname>.<lastname>@<anymail>.<fld> :-((12:44
CcSsNETtrollathon in here this morning12:44
CcSsNETbye for now12:44
wolfspraulCcSsNET: yes, I said it's a distraction12:44
methriltrollathon? we are just chating12:45
DocScrutinizerindeed, an offensive bold comment12:45
wolfspraulCcSsNET: wanna introduce yourself? the rest here all know each other...12:45
wolfspraulwhat takes you here, what interests you?12:46
wolfspraulDocScrutinizer: I hope I can take you up on some Milkymist KiCad action :-)12:46
wolfspraulgive it another week or so and we start...12:46
DocScrutinizerlet's see what my plans to get fresh PC with fresh linux will result in12:47
CcSsNETo so yall know each other. hmm12:47
methrili need to found time12:47
methriland read my FOSS e-mails again12:47
wolfspraulit should still fit in your free space, in fact it's easy to build from source even12:47
DocScrutinizerCcSsNET: especially we are a nice and friendly bunch in here, accusing us for trolling is a bit rude12:48
DocScrutinizerwhen wolfspraul or any other of the well known all time menver would do that, it might be OK. You shouldn't though12:49
wolfspraulCcSsNET: he, thanks! [your site]12:49
wolfspraulquite opinionated :-)12:49
CcSsNETread my twitter12:50
CcSsNETi talk lots of shit there directly at companies12:50
wpwraknickoe: fped has rectangular and half-circle + rectangle + half-circle pads12:50
DocScrutinizermoo wpwrak 12:51
DocScrutinizerseen my mail? :-D12:51
wolfspraulgood morning Buenos Aires12:51
DocScrutinizerwpwrak: crowbar sync: bom +=312:52
wpwraknickoe: the latter also include circular pads (if the rectangle inside is zero and you just get two half circles)12:52
methrilhi wpwrak!!12:53
DocScrutinizerplus additional benefir on top, for free :-)12:54
wpwraknickoe: you can disable the overlapping pads check by editing the .fpd file: put  allow touch  or  allow overlap  after the "unit" directive12:54
DocScrutinizertotal bom: 2 Zener, 2 diodes, 1 thyristor, 2 R, 1 C (from top of my mind)12:54
wpwrakDocScrutinizer: no mail ... where did you send it ?12:55
DocScrutinizerreply all to yours12:55
DocScrutinizerto ML CC: wpwrak12:55
wolfspraulhow much space does all this occupy?12:57
Action: wpwrak checks his spam folder12:57
DocScrutinizermax 3cm^212:57
DocScrutinizerprobably 1.512:57
wolfspraulwow that's huge12:57
DocScrutinizererr 212:57
wpwrakDocScrutinizer: not a trace of your mail anywhere :-(12:57
wolfspraulyou mean 2cm by 2cm ?12:57
DocScrutinizer2 cm^212:58
DocScrutinizernot (2cm)^212:58
wpwraknickoe: but disallowing the checks is usually not a great idea. you may need it for things like PCB antennas, though12:58
DocScrutinizerM1 didn't look like it's crowded12:59
wolfspraulstill a lot I think. don't even know about cost, but size-wise, a few extra 0402/0603 components are fine, but anything bigger than that makes me wonder whether it's the right path12:59
wpwrakmethril: let's see how long until you get bored of .ie and return to the south ;-)12:59
methrilwpwrak: let's see13:00
wpwrakDocScrutinizer: it well be once we're done with r4 ;-)13:00
DocScrutinizerit's not enymore after I weeded it out ;-D13:00
DocScrutinizerrecall the surplus audio amp in GTA02? ;)13:01
wpwrakah well .. :)13:01
DocScrutinizerOM said "thanks for finding that, but no way you got some 'own 1cm^2 now to push in other stuff you like"13:02
DocScrutinizeryou still owe me that 1cm^2 ;-P13:02
DocScrutinizeranyway, real estate needed mostly depends on size of sureproof thyristor and diodes13:04
DocScrutinizerwhich in turn depends on buffer C size you got on that board13:04
DocScrutinizerand of course on sourcing a nice set of components that can do high non-repetitive surge on small package13:05
wpwrakmaybe you should check your mail queue. if you really sent it to the list and to me, then at least something ought to have arrived13:06
DocScrutinizer(buffer C) not only size but also parameters like ESR etc13:06
wpwrakif you're not subscribed to the list, wolfgang will have to add an exception for you, though13:06
DocScrutinizerI'm not subscribed13:07
DocScrutinizerbut CC should've arrived13:07
DocScrutinizerwell, maybe it's still on the way, I sent it 90min ago13:08
wpwrakdeutsche post ? :)13:08
DocScrutinizerOM mailer13:08
DocScrutinizerI'll resend to you, from another account13:09
DocScrutinizermeh going to spam it here - prepare for impact13:10
DocScrutinizerbtw I just thought connecting both VDD rails (3V3 and 5V) to the thyristor via some burst-proof diodes (10A or sth, maybe Schottky), and getting a 2.1V Zener form 5V to 3V3, would not only solve the sync problem (OverVoltage on 5V triggers 3V3 OVP detection via Zener. Then both get crowbarred via their own diode and common thyristor), but also ensure a more sane power-up sequence (Zener working as diode when biased reverse makes sure 13:10
DocScrutinizer5V rail never is significantly lower than 3V3, forward Zener ensures 3V3 comes up in sync with 5V)13:10
wolfspraulCcSsNET: welcome to Qi Hardware first of all, I don't recall you here earlier13:13
wpwrakah, mail from your grandmother's project arrived :)13:13
wolfspraulthanks for stopping by, and feel free! :-) some people here have several year history, but don't let that make you feel it's not your place somehow. the only thing we really care is to be as friendly as possible.13:13
wpwrakDocScrutinizer: hmm, couldn't this cause trouble when powering up ?13:14
wolfspraulwhich sometimes collides with our passion, admittedly :-)13:14
DocScrutinizeropen mobile, made available ;-P13:14
wpwraki'm also not quite sure how you're start the thyristor. a schematic drawing would help :)13:15
DocScrutinizerwpwrak: which kind of trouble? only problem I see is power on Zener13:15
DocScrutinizerit's on the contrary avoiding some otherwise possible problems13:15
wpwraktrouble = on power up, the 5 V rail is up before 3.3 V. so if this trips the crowbar ...13:15
DocScrutinizerthis doesn't trip the crowbar, and 5V up befor 3V3 is cured by Zener anyway13:16
DocScrutinizerZener might fuse when 3V3 never comes up though13:17
DocScrutinizere.g due to a short on 3V3 rail13:18
DocScrutinizeras Zeners always "fuse to short" </quote>, this would cause both 3V3 and 5V either engage short protection, or (after short removal) cause crowbar 3V§ OVP trip by the short to 5Vintroduced by short Zener13:19
DocScrutinizerdepending on 5V regulator I'd expect both rails go into overcurrent protection with a short on 3V3 though13:20
DocScrutinizerand Zener survives13:20
DocScrutinizeronly "risk": make sure 3V3 has some minimal load *always* as otherwise the Zener leakage might cause OV on 3V3 and trip crowbar13:22
wpwraki think we have that13:22
DocScrutinizerI'm sure about that :-)13:22
DocScrutinizerI think it's a pretty simple yet nifty and failproof design for a OVP13:23
wpwrakhm. i'd be worried about obscure failure modes13:24
DocScrutinizerplus en passant makes sure no too nasty things happen on power-up sequencing13:24
DocScrutinizernah, I already evaluated all whatever obscure failure patterns13:25
DocScrutinizerit's safe in my book13:25
DocScrutinizerrock solid design13:25
DocScrutinizerno matter which component fails in which way13:25
DocScrutinizerof course there'S no substitute for tests on real iron13:25
wpwrakthe other thing i don't like about it is that we can probably get all the OVP we want by just picking the right regulator chips. so once we finally get to clean up the power circuit, which is a mess, then we should be able to have a nice integrated solution (where someone else has already figured out and fixed the weird corner cases :)13:26
DocScrutinizerno regulators protect against reverse fed OV13:26
wpwrak(power up) externally regulated 5V is the main supply for everything. 3.3V is derived from 5V13:27
DocScrutinizerexcept shunt reulators13:27
DocScrutinizerplus a crowbar is meant to catch component brakage in rgulator block as well as externally caused OV13:28
DocScrutinizerso you'll hardly find a regulator chip with built-in crowbar13:28
DocScrutinizer(shunt) deutsch: quer-regler (as opposed to the usual laengs-regler)13:29
DocScrutinizerquer-regler not commonly used anymore, since they have abysmal efficiency13:30
DocScrutinizeronly place to use them is to regulate 20+kW PSU when load changes a few 2 or 3% only13:31
DocScrutinizerthen your parallel/shunt regulator only consumes some 600W rather than several 100013:32
wpwrakmmh. nice and warm in winter :)13:33
DocScrutinizerduh, actually each Zener stabilization is a shunt regulator circuit13:33
DocScrutinizerthe one BIG advantage of shunt reg: extremly low source impedances are possible for the PSU13:35
DocScrutinizeras current from buffer/source has no huge transistor or FET to go thru13:35
DocScrutinizerowadays with rather low R(DS) FETs this argument is somewhat moot, it was valid intimes of germanium transistors13:36
DocScrutinizernow we're *really* OT :-P13:37
wpwrakjust checked the 3.3 V regulator we use. doesn't do anything fancy, which was to be expected13:39
DocScrutinizerof course not13:39
DocScrutinizerfor OVP you won't find anything reasonable on regulator domain, as you also got buffer C (a lot of them)13:40
wpwrakbtw, an integrated solution can also simply consist of OVP detection with a global shutdown. (e.g., with a multi-voltage regulator)13:40
DocScrutinizerOVP needs to discharge all those buffer C in NO time -> crowbar13:40
wpwrakyeah, sounds reasonable13:41
DocScrutinizerwhich effectively answered your last post before it got posted13:41
wpwrakbut ... post-M1r4, i think. adam had enough fun with the even simpler input protection circuit of M1rc3. i don't think he'll appreciate getting a crowbar added to all the other changes in M1r4, of which there are many13:42
DocScrutinizerof course you can try to find a better BOM /circuit for the crowbar, but you won't get much better on it than the prev quoted BOM13:43
larscif this was twitter "crowbar" would probably be the trending word for the day13:44
DocScrutinizerif he's not willing to add a circuit that simply can get NC by not populating the components, he does something wrong13:44
DocScrutinizernot A SINGLE bridge/0R needed, just don't populate the components - can you come up with a design that has less risk to implement on next production rev?13:45
DocScrutinizerif adam refuses to consider it, you should find somebody else for that task13:46
nickoewpwrak, hi, I just got back. But how do I make half-circle pads? There is no GUI button to do that.13:51
DocScrutinizerit's not like I'd be more daring on design changes than any of you, au contraire. It's just I have (like with all of my code) already a catch-all/exit-strategy for *every* possible oopsie when I come up with a suggestion13:52
wpwraknickoe: the rounded pads consist of half-circle plus a rectangle (may be zero) plus another half-circle13:53
nickoewpwrak, but in the file it just says rpad for round pads13:54
wpwraknickoe: the half-circles are internal drawing elements (and the whole pad becomes an "oval" pad in kicad anyway)13:54
wpwraknickoe: yes. round = circle or rectangle terminating in half-circles13:55
nickoemaybe I should ask you if fped can construct a pad... I will illustrate.13:56
DocScrutinizerand I hate the approach of "we got 3 bugs to fix. Let's begin with first one on rev after next, and see how it goes"13:56
wpwrakDocScrutinizer: well, if adam is keen on adding this, i won't stop him ;) but i'm also concerned about the overall workload13:56
DocScrutinizerthis will slow down things to a grinding halt, and cause 27 hw rev13:56
wpwrakDocScrutinizer: don't forget that the size of our army is rather small13:56
DocScrutinizeroh hell, I probably could add that crowbar in an afternoon's work, even when I also had to install and learn to use kicad same time13:57
DocScrutinizererr, sorry, scratch that - I forgot sourcing13:58
nickoewpwrak, I was thinking of drawing something like: http://kom.aau.dk/~nickoe/pad.svg13:59
DocScrutinizerso my approach probably would be to get footprinta for what I think are rather standard packages for the components we need, then deal with it later, after the pressing work got done13:59
DocScrutinizeradding the footprints for standard components is actually a task for a saturday evening, between sauerbraten and disco14:00
DocScrutinizerand if those footprints go unpopulated on *all* sales boards, so what? I still got a nice eval biard from standard production to test the crowbar circuit14:01
DocScrutinizermy old rant since before I joined OM: footprints are for free14:02
nickoeDocScrutinizer, what is OM? OpenMoko?14:02
DocScrutinizeronly resource needed and used up: your brain14:02
DocScrutinizerwhat else?14:03
DocScrutinizer(of course footprints also use the resource real estate, which often is the limiting factor)14:03
nickoedunno, but what exactly happened to OM and gta02-core?14:03
larscDocScrutinizer: "oblivious monkey" ;)14:04
DocScrutinizeryou should ask wpwrak about gta02-core14:04
nickoewpwrak, I have to go for some time now, but I will be hanging around, so just msg me, if you intend to answer my question about, how to draw the pads, http://kom.aau.dk/~nickoe/pad.svg.14:05
DocScrutinizer...and wolfspraul about "what happened to OM"14:05
wpwraknickoe: interesting shape :-) how about keeping it very simple, with a 0603 or similar, with a trace between the pads ?14:07
DocScrutinizercould've been my sentence, though I got no context14:07
wpwraknickoe: you could also use two square pads instead of the ones with half-circles and have another pad in the middle. but you'd have to  allow touch  for that14:08
DocScrutinizerNormalyClosed bridge, to cut open?14:08
wpwraknickoe: i'd go for the 0603/similar footprint ;) that one can also be easily reworked14:09
wpwrak(gta02-core) it nodded off and never woke up again ...14:09
DocScrutinizerI favourize two rectangular pads with a 0nly 0.5mm gap between them so you can easily bridge by a solder blob14:10
DocScrutinizertrace to cut goes to the side of gap, not in between14:10
DocScrutinizerprobably I'm again OT now, due to missing context14:12
wpwraknickoe: with fped, if you  allow overlap  you could also use a round pad and overlap one side with a rectangular pad, but i would expect this to cause trouble down the road, either in kicad, or at the pcb maker14:13
DocScrutinizeranyway on that pad http://kom.aau.dk/~nickoe/pad.svg cutting the trace is horribly awkward14:14
wpwrakyou'd also have to do the solder mask layer manually in this case14:14
wpwraktwo cuts, left and right, and lift off what's between ?14:14
DocScrutinizerbetter use sth shaped like °U°14:14
wpwrakor solder a 0R ;-)14:15
DocScrutinizerso you can cut the trace between the two ° pads at "foot" of "U"14:15
wpwrakbonus: you can even run a trace under it, if necessary :)14:15
DocScrutinizerindeed, which is about the only reasoning to get 0R actually14:16
wpwrakplus, you don't need to take a knife to your pcb :)14:16
DocScrutinizerwhen there's no trace to bridge, you frequently ask "why the heck a 0R and not simply a trace to cut with just footprints for a 0R?"14:17
wpwrak"child-safe pcb design" ;-)14:17
DocScrutinizerof course you should keep away other traces from the point where you plan to cut14:18
wpwrakah well. de gustibus ...14:18
DocScrutinizereach 0R is a component that can go awry on soldering14:18
wpwrakone advantage of 0R is that it's an SMT option, if you decide to unuse it permanently14:19
DocScrutinizerand honestly: how often did we remove any of those silly 0R at OM? How often we would've been thankful for it being a component to unsolder rather than a trace to cut?14:19
DocScrutinizersure, decide to not use V-CORE1V2 in GTA02 XP14:20
Action: DocScrutinizer quits sarcasm mode, sorry14:21
DocScrutinizerI completely agree with usecase of opt-in/out "jumpers" for 0R14:23
DocScrutinizer"make X NC, make Rx 0R" / "use X, make Rx NC"14:24
DocScrutinizereverybody does that14:25
DocScrutinizerbut for "probing current from LVDO1"? c'mon14:25
DocScrutinizerthat's pototype design cargo-cult-prted to production boards14:26
DocScrutinizerwhere we are at crux of the issue: all layout and EE design in OM been cargo cult driven14:27
wpwraki thought it was kaka-cult ;-)14:29
DocScrutinizernah, C&P and cargo-cult14:29
DocScrutinizerspiced with a good dash of cluelesness where needed14:30
DocScrutinizerhooking up signals based on the naming, no matter if the actual semantics and levels match or not14:31
DocScrutinizerrouting completely useless signals from meaningless subsystems to a GPIO so maybe eventually somebody might use it (or not, and if then for what exactly?)14:32
wpwraki don't think we had a lot of these14:32
wpwrakperhaps one or two on gps14:33
DocScrutinizerwe got that one overcurrent signal from that switch on a GPIO, which allowed us to diagnose a broken PCF50633 VUSB function ;-P14:33
wpwrakah yes. "help ! my power supply has failed !" ;-)14:33
DocScrutinizeractually it would've been a good thing if we ever switched to a switch with integrated OVP there, according ro my suggestion during dispute with Andy about USB OV14:34
DocScrutinizerdetectinf OC is a meaningless thing in this context14:35
DocScrutinizerjust the switch had that OC signal, so they routed it to a GPIO. Period14:35
DocScrutinizerbtw basically the whole switch was surplus14:36
DocScrutinizeras the PCF50633 had its own way to switch this input14:36
DocScrutinizerbut then it's been this very switch that gave us some awesome xray microscopic photos about burnt-out die14:37
DocScrutinizerwhich by itself would have been the killer argument for getting a OVP protection on USB, alas Andy didn't appreciate that14:38
wpwrakbtw, is your grandmother project a public address ? or shall i replace it with OminateMonkeys when forwarding your mail to the list ?14:39
DocScrutinizerand hell, the switch even was 6V ABSMAX while PCF50633 only was 5.514:39
DocScrutinizerno public addr14:40
DocScrutinizermy raw work mail acct14:40
DocScrutinizerno proper spam filtering there14:41
Action: wpwrak opens a fakebook account :)14:42
Action: DocScrutinizer warms up the MTHELs14:42
DocScrutinizeralready been time to evaporate some doggies shiting on my lawn14:43
wpwraki could use some for the doggies barking in the neighbourhood14:43
wpwraklike, incessantly. every day. for weeks.14:44
DocScrutinizerrun for shelter, I'm spending one of my geostationary nuklear powered Roentgen Lasers to cure your problem14:44
DocScrutinizermayybe diving in your bathtub might block the secondary radiation (only if targeting precisely) ;-)14:45
wpwrakhm. i was actually just thinking of taking a shower. must be my x-ray instinct :)14:47
DocScrutinizerwpwrak: calculation shown the impact area is a bit large for one backyard and also has not that precisely defined a border, also I the heck can't decide what to do with the other 15 lead rods of that particular thermonuclear powered xray laser device, and finally I feel unconfortable with polluting the orbit with nuclear waste. Maybe use some rat poison instead, hidden in saussage?15:03
DocScrutinizernot even taken into acccount possible impact on TV sats, OMG :-o15:07
Action: DocScrutinizer triggers another calculation about radiation level expected at next nearby TV sats - just in case to know it for the future15:08
DocScrutinizerpabs3: thanks for answering that mail regarding xchat on maemo, I actually been a bit lazy to answer all those questions and I think you covered the best part of it15:17
DocScrutinizerhell no, maemo is NOT a x86 system ;-P15:18
pabs3DocScrutinizer: actually there is one email they asked you some questions in but forgot to add you to the CC. I will bounce it to you now15:18
DocScrutinizerthnaks :-/15:18
pabs3DocScrutinizer: basically they haven't been able to reproduce it15:19
DocScrutinizerI however was able to reproduce it, exactly the way they described it for win-xchat15:20
pabs3on Linux?15:20
DocScrutinizeron maemo15:20
pabs3you got the mail?15:21
DocScrutinizernot yet15:21
DocScrutinizerI even provided a cleanup scriptie, don't you see it?15:22
pabs3I did15:22
pabs3I also tried to reproduce it but couldn't on Debian amd6415:23
DocScrutinizerwell, the mail you forwarded already arrived me before and I referred to it and your answer on it in my post above15:24
DocScrutinizeryes, I talked about armel version, not x86 version. Yes it was reproducable exactly the way they analyzed for win-xchat in several blogs/bugtraq/ML, and no I dunno a shit about which GTK version is running on maemo15:25
DocScrutinizerand the rogue string got sent to me in #maemo15:27
DocScrutinizerwhich instantly and permanently (until backsroll cleanup via my script) segfaulted maemo based xchat15:27
DocScrutinizerso I'm not going "to reproduce it" again for academic reasons, it's been real15:29
DocScrutinizerif you want me to run it under strace/gdb or sth, I might reproduce by sending the rogue string to myself via IRC. But I'm really reluctant to do any of that unless those maintainers exhibit a bit more of getting what I already said15:31
DocScrutinizer(you're free to quote me, though)15:32
pabs3will forward this to them15:32
kristianpaul2SMZ sounds like a xtal reference?15:37
kristianpaulah yes quartz15:40
Action: nickoe wpwrak, I am back, and readding log17:35
nickoe"sure, decide to not use V-CORE1V2 in GTA02 XP" eh17:37
nickoewhat does that mean?17:37
qwebirc68040hi all18:08
qwebirc68040I am using ubuntu 11.10 kernel 3.018:08
qwebirc68040I cannot connect to ben nananote18:08
qwebirc68040by usb18:08
qwebirc68040the inferface works18:08
qwebirc68040but it has not the ip indicated by wiki18:09
DocScrutinizernickoe: It meant "there's no use in having a 0R that allows to to switch off one of the main powerrails of your device"18:09
nickoeahh, okay, thats is a point18:10
qwebirc68040this is the output18:11
qwebirc68040usb0      Link encap:Ethernet  HWaddr 62:07:05:67:16:92             indirizzo inet6: fe80::6007:5ff:fe67:1692/64 Scope:Link           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1           RX packets:0 errors:0 dropped:0 overruns:0 frame:0           TX packets:47 errors:0 dropped:0 overruns:0 carrier:0           collisioni:0 txqueuelen:1000            Byte RX:0 (0.0 B)  Byte TX:12658 (12.6 KB)18:11
qwebirc68040it is different than the one I used to have with the other versions of ubuntu18:12
wolfspraulI wouldn't follow the "remove connection manager on Ubuntu" part though, sounds a bit radical18:14
wolfspraultry: ifconfig usb0
wolfspraulthen ping (with ben connected)18:15
qwebirc68040I will do that18:15
wolfspraulqwebirc68040: btw, even though you have trouble right now - thanks for getting a Ben and coming here to ask for help :-)18:16
qwebirc68040thanks now it works18:16
qwebirc68040please update the wiki with this new info about ubuntu 1118:17
wolfspraulwhat is the new info?18:17
wolfspraulwhat I told you came straight from the wiki18:17
wolfspraul(just that I don't like the idea of removing the connection manager, but I would need to do some more tests before updating the wiki page)18:18
qwebirc68040sorry with ubuntu I had not to set up the ip with comman line18:19
qwebirc68040command line18:19
qwebirc68040I would use a command line version of stardict18:22
qwebirc68040I think is called sdvc18:22
qwebirc68040Is there a package for this?18:22
wolfspraulmaybe it's in the default image already?18:23
qwebirc68040sdvc command not found18:24
wolfspraulhmm, packages are here http://downloads.qi-hardware.com/software/packages/NanoNote/Ben/latest/18:24
wolfspraulbut no sdvc there either, at first glance18:24
wolfspraulwe need to ask xiangfu about sdvc tomorrow18:25
qwebirc68040and to use cyrillic characters I remember it was necessary the ben-cyrillic package. Am I right?18:26
wolfspraulfor that we have some experts here too18:28
wolfspraulhave to wait until they respond though18:28
wolfspraulI would think it's preinstalled, but I could be wrong18:28
qwebirc68040I would use gftp to tranfer to ben some opkg files 18:35
qwebirc68040with ftp what port should I use_18:36
qi-bot[commit] Lars-Peter Clausen: ASoC: jz4740-pcm: Fix snd_pcm_hw_constraint_integer error check (jz-3.2) http://qi-hw.com/p/qi-kernel/db41fa019:31
qwebirc73397hi I installed the ben cyrillic package 19:35
qwebirc73397but I do not remember how to activate the russian characters .. perhaps with the ben .. chinese button19:36
kristianpaulargh, why latex2pdf need so much space and packages !21:56
kristianpaulwhat??!!! texlive-latex-extra 358Mb !!!21:57
kristianpaulno way21:57
wpwrakthere's a whole little ecosystem in there ;-)22:05
wpwrakDocScrutinizer: quick sanity check: given a ribbon cable with signals next to each other, no ground between signals, to make the best of this messy situation, i'd recommend:22:35
wpwrak1) keeping the cable short, and 2) keeping signal edges slow. does that sound like reasonable advice ?22:35
Jay7KDE+Mer tablet22:43
Jay7so.. it's time to run with Ya.tablet ;)22:45
wpwraknice :)22:45
DocScrutinizerwpwrak: yes22:48
wpwrakthanks ! in a ~10 MHz system, with how long a cable would you estimate one would run into problems ? 10-20 cm still tolerable ?22:50
wpwrakat what point would it start to get nasty ?22:50
DocScrutinizerof course depends on source impedance etc, but generally no problem22:51
DocScrutinizerI'd frown at 50cm22:51
wpwrakokay, so you wouldn't go much further then22:51
wpwrakah, quite a bit22:51
Aylait depends of the frequency22:51
wpwrakvery good. thanks !22:51
DocScrutinizeryou *might* get meters without problems, as mentioned it depends on a nuumber of factors22:51
wpwrakyup. i'm assuming ~ 10 MHz. my be a tad more.22:51
AylaI remember I did learn those formulas. Don't remember any :)22:52
wpwraksure. i just want a ballpark number. like "a few 10 cm is probably safe" vs. "5 mm MAXIMUM !" ;-)22:52
kristianpaulDocScrutinizer: what about same wpwrak questions but onw 100 Mhz :-)22:52
wpwrakkristianpaul: are your I/Os really running at 100 MHz ?22:53
DocScrutinizeras a giudeline: I seem to remember ISA bus extenders that had 30cm of cable22:53
kristianpauljust curious :-)22:53
wpwrakokay, ISA was DC ;-)22:53
DocScrutinizerkristianpaul: 100MHz will possibly fail on <10cm, in my book22:54
DocScrutinizermight even fail on 2cm, when you spoil things22:54
wpwrakyeah, i think you'd be happy if the signal makes it through the connector ;-)22:55
Aylathe wavelength of a 10MHz wave is 30m22:55
wpwrakwe already have a bit of traces on the main pcb, then some 15 mm of connector, etc.22:55
DocScrutinizerwavelength is irrelevant22:56
kristianpaulAyla: but you usualli get a IF for it no?22:56
DocScrutinizerit's the impedance matching that counts22:56
kristianpaulAyla: and you can decimate as well ;)22:56
kristianpaulIntermediate Frequency22:56
DocScrutinizeralso you don't worry about crosstalk when the signalling is stable22:57
DocScrutinizerI.E. NRZ etc22:57
DocScrutinizerproper levels22:57
DocScrutinizerUSB using 480MHz22:57
DocScrutinizerand you get cable length of >1m with good cable22:58
DocScrutinizeror see 100BT22:58
DocScrutinizerrule: NEVER get sharp bends into the cat5a cable22:59
kristianpaulmy M1's usd cable is 1.2m :-)22:59
DocScrutinizeragain: impedance22:59
kristianpaulwhat about LVDS, when we really need it?23:00
kristianpauli tought, for example a IF Tunner and a ADC, uses LVSD, why??23:00
DocScrutinizerLVDS should be immanently ruggedized signalling23:01
kristianpaulthat sounds good :-)23:01
DocScrutinizerthe whole trick of all these is each time the differential signalling23:02
DocScrutinizera twisted pair behaves rather neutral to environment when driven balanced23:02
DocScrutinizerUSB: D+/D-23:02
DocScrutinizercat5a 1GBT, same twisted pairs23:03
DocScrutinizerLVDS: also differential aka balanced23:03
DocScrutinizerwpwrak though asked about a flat ribbon cable without GND separator wires23:04
DocScrutinizerand that's a bit more of a problem23:04
DocScrutinizerand flat ribbon connectors usually are post type, which isn't known for excellent wave-impedance matching23:05
kristianpaul"GND separator wires" is a shielding?23:05
DocScrutinizerkind of23:05
DocScrutinizerusually you have signals on one side of the 2 row pin connector, and GND on the other one, which results in signal, GND, signal, GND sequence of wires in flat cable23:06
wpwrakkristianpaul: you normally have GND - signal A - GND - signal B - GND - ...23:06
wpwrakkristianpaul: that is, on the cable23:06
DocScrutinizerexactly :-D23:06
kristianpaulha, yes i tought that23:06
kristianpaulwhere i get that cable? :-)23:07
wpwrake.g., IDE cable are like that23:07
wpwrakit's not the cable that's special. it's the pin assignment23:07
kristianpauland wait, but i still need that GND grouding on the PCB for each extra GND wire?23:07
wpwraksee my mail ;-)23:07
kristianpaulahh i just checked 5m ago ;)23:07
wpwrakDocScrutinizer: btw, you really ought to subscribe to the milkymist list. even if you don't read it, we could then at least simply tell you to check when there's something that's relevant for you.23:08
kristianpaul"extension board " yup also cables are messy on  my desk..23:08
wpwrakDocScrutinizer: this would allow you to post replies without getting rejected, breaking the thread, etc.23:08
DocScrutinizermeh, toss a pipermail URL ;-D23:08
DocScrutinizerno way23:09
kristianpaulah you did one J21 exp board for leds..23:09
DocScrutinizerI already disabled my email display in systray, as it didn't manage to cope with 5 digits ;-P23:10
wpwrakkristianpaul: of course :)23:10
wpwrakkristianpaul: using whatever connectors i had sitting around :)23:10
kristianpaulhad you seems my EVB connector pich? :)23:11
wpwrakwell, i could probably also have used a proper Nx2 critter but the i would have had to add a cable23:11
wpwrakhmm, not sure. maybe show it again :)23:11
DocScrutinizerGPS? "we" just have another deja-vu on GTA04 with GPS being poisoned by device's EMI radiation XD23:14
wpwrakkristianpaul: have you had a fix yet ? ;)23:15
DocScrutinizerexpect your cable to radiate like a tesla transformer, when it comes to non-GND-protected unshielded flat ribbon cable with 10MHz signals on it23:15
DocScrutinizeras soon as GPS is involved which deal swith S/N ratios that are a nightmare basically23:16
kristianpaulwpwrak: EVB http://wikisend.com/download/237672/top_side.jpg23:16
DocScrutinizerthere are even shielded twisted-pair flat ribbon cables23:18
DocScrutinizeriirc used for SCSI-323:18
wpwrakDocScrutinizer: like this ? ;-) http://2.bp.blogspot.com/_NVQujPOPx2c/TDiRmBNrQFI/AAAAAAAARQs/hqb1PwdJYmk/s400/2.jpg23:19
DocScrutinizeryeah :-P23:20
wpwrakkristianpaul: ah, that's just the EVB. so you simply connect to the header23:21
DocScrutinizerI'd not expect to get a GPS fix with antenna <1m away from M1 with extension cable, resp away from cable (any part of it)23:21
kristianpaulheader is not same pitch23:21
kristianpaulDocScrutinizer: oh sure23:21
kristianpaulDocScrutinizer: is 3 meters out to the window23:21
wpwrakout the window, then a cable over to the friends at arecibo ...23:22
DocScrutinizerso what's the problem with that cable, kristianpaul ?23:23
DocScrutinizeryou already are in trouble with it?23:23
DocScrutinizeraah, ok23:23
kristianpaulnope that i know, just wpwrak made me think about it yet23:24
DocScrutinizerI'd first see if any troubles arise23:24
DocScrutinizer10MHz still is kinda LF23:24
kristianpaul16Mhz actually .)23:24
DocScrutinizerwhat signals do you transmit there?23:25
DocScrutinizertiming critical?23:25
kristianpaulIF sampling real data two bits23:25
DocScrutinizer2bit PCM?23:25
DocScrutinizerwith clock?23:25
kristianpaul16.384 Mhz23:25
DocScrutinizerwith, not which23:26
DocScrutinizerI.E. sync or async23:26
DocScrutinizermake sure you get proper termination at both ends23:27
kristianpaulproper :/23:27
DocScrutinizerconsider clock line timing adjustments by introducing delay lines to get some 90° phase shift23:28
DocScrutinizerso your data lines are sampled in center of eye-pattern23:28
kristianpaulDocScrutinizer: http://en.qi-hardware.com/w/images/2/2a/Sigelogic.png 1. clock 2. sign 3. magnitude thats what came trought ribbon cable23:28
DocScrutinizeraah your clock is twice the data rate? good23:29
DocScrutinizerthough, WTF are those discontinuities on CLK?23:30
kristianpaulcheap logic analizer23:30
kristianpaulin scope looks better23:30
DocScrutinizerso I gather you sample on rising edge of clock?23:32
kristianpauli got same discontinuities with 4Mhz clock..23:32
DocScrutinizerthe discontinuities look like aliasing between time frame pace of logic analyser and real signal/clock23:34
DocScrutinizerI gather your analyzer just does 2 or 3 fold oversampling23:34
larscor not even that23:35
kristianpaulclock is twice the data rate, no, data rate is same as clock23:36
DocScrutinizeractually sligtly <223:36
larsclooks like 30Mhz or something23:36
kristianpaulbut i still not worry as i decimate later23:36
kristianpaulalso all the SoC that sample this signals (milkymist) run syncronous from the same gps receiver clock23:37
DocScrutinizerjust make sure you sample in centwer of eye pattern on RX side23:37
kristianpaulalso tha pic above is not my current sampling setup (sorry i was trying to give an idea)23:38
DocScrutinizerdon't forget you got quite some propagation delay on a 50cm flat ribbon cable, and that this delay might differ between clock and data lines, if only a bit23:39
kristianpaulribbon is ~13cm23:40
kristianpauli'll move to PCB-glue solution later anyway23:40
DocScrutinizeryou might want to sync a PLL on RX side and basically run async23:40
DocScrutinizerrecovering clock from edges on datalines as well as clock23:41
kristianpaulin the other side (FPGA) there is a buffer23:41
DocScrutinizers/as clock/as clock line/23:42
DocScrutinizerbut maybe that's all a big overkill, and you're just fine with what you got there23:43
kristianpaulfine for now :)23:43
kristianpaulas wpwrak said i still not have a fix23:43
kristianpaulalso i dont need it yet, as soon as i got tracking a satellite :)23:44
kristianpaulDocScrutinizer: but i already have your advices, wich is relly great23:45
kristianpaulconsidering the atenna will come later to the same box, i need think about that too23:45
DocScrutinizerI'm always glad to share some of my... ideas23:45
kristianpaulI'm open to listen it :)23:46
DocScrutinizermeh, I should go to sleep23:47
kristianpaulgood sleep then :)23:47
kristianpaul18:48 still here and still lot to do..23:48
DocScrutinizersleep $(( 60 * 60 * 4 ))23:48
kristianpauland drive to work? :-)23:49
DocScrutinizerhttp://www.stericsson.com/products/m5730-thor.jsp  well23:57
--- Mon Jan 30 201200:00

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