#qi-hardware IRC log for Thursday, 2010-09-02

wpwrake.g., i went through three of them for the blinkenlights. the one on the video and pictures is a first version board, but the final one has better mechanical properties. didn't actually complete the board, though. but i'll use the data for ben-wpan00:00
wpwrakkristianpaul: tx power seems to make zero difference :) i set it to 251 mW and dumped some files at the PC next to my TV over wifi. so either that pc sends with more power than the access point or the tx power setting doesn't matter much00:01
wpwrakwolfspraul: btw, the cnc mill is an important bit of the rapid r&d setup. not only can i do messy shapes with it, but it also does that in a fairly reproducible way.00:03
wpwraki still need to connect board outline in kicad with heekscad, though. right now, i manually write a script that generates a toolpath with my own set of utilities. (the stuff in cam/, mainly pcb.pl)00:05
wolfspraulwpwrak: you mean for the shape of the PCB?00:08
wpwrakkristianpaul: well ... poking around a bit more, i see a bit of a difference. peaks are the same, but the average changes when i increase tx power. so maybe the pc's acks are just very strong.00:08
wpwrakwolfspraul: yes. e.g., idbg and blinkenlights both have very precise shape specifications. ben-wpan is a bit more relaxed for now, but that won't last for long00:09
wpwrakhmm, bit-banging on and off, with a ben, only 7 MHz. a bit disappointing.00:44
wpwrakwhile (1) { PDDATS = 1 << 9; PDDATC = 1 << 9; }   compiled with -O900:46
rohcan beat that with a sbi cbi loop on avr00:46
rohat 20mhz00:46
rohwell.. its not intended as a bitpattern-sw-generator.00:47
wpwrakstill ...00:47
wpwrakthe loop gets unrolled a little. twice. besides that, it's just the store instructions and the jump00:49
wpwrakbut the jump is probably absorbed by write buffers anyway. at least i don't see a timing difference on the scope.00:51
wpwrakbut now to the fun part ... outputting the sd/mmc clock00:52
rohunroll it more.. like 10 times. who knows what the cache latency is01:08
rohjust to see if it makes a difference01:08
wpwraknaw, that doesn't make sense. i should be hammering it already with something like 30 MHz or more01:10
wpwrakand there we are. a nice 16 MHz for my transceiver. bye bye crystal :)01:38
larscwith the mmc clk?01:51
wolfspraulwpwrak: reading the backlog. what do you mean with 'sd card friction lock' (so the card can be 'flush with the edge')?01:55
wpwrakwhen you insert the uSD card, you have to push it in a little to it "clicks" into place01:56
wpwrakit then also comes back out a bit01:56
wpwrakfor an internal card, that's probably undesirable. so instead of this push-to-insert and push-to-release mechanism, it may be better to have a mechanism that holds the card just by friction01:57
wolfspraulwhy is it undesirable?01:57
wpwraknot sure if such card holders exist, though. a quick search didn't find any01:57
wolfspraulwhat do you mean with 'internal card'?01:57
bartbeswpwrak: well, they are in some mobile phones, so I'd guess they exist01:58
wpwrak(internal card) well, if uSD works great for extension boards, why not have one on the inside as well ?01:58
bartbeswhether they're for sale is something entirely different of course01:58
wolfspraulyes sure01:59
wpwrakbartbes: naw, with this sort of thing there are no crazy limitations01:59
wolfspraulbut I don't understand what the problem is with the little 'push in' to release the card01:59
wolfspraulyou want a solution mechanically that essentially has no 'lock' at all. you just push it in to a certain point, and from that point you can pull it straight back out the same way it came.02:00
wpwrak(undesirable) wastes space, may be less stable (your lever grows), etc.02:00
rohwpwrak 'great for extension boards' .. well.. you have a funny view on things sometimes...  i find the mechanics rather tricky to do myself. and i got a cnc mill02:00
wpwrakat least i think that may be better02:00
wolfspraulwhy does that waste less space?02:00
wolfspraulbecause you don't have to reserve a little room before you can have thicker components?02:00
wolfspraul(that space would be necessary for the push-in-distance)02:01
wpwrakif the card has a width W, then you need W+travel space for having it installed with that mechanism02:01
wpwrakyes, on the card as well. you can view if from either side02:01
wolfspraulthat push-in is at most 1mm or so02:01
wpwraksure, it's not a lot. but if you card is maybe 10 mm ...02:02
wolfspraulI think the idea is that it allows you to insert microSD cards that are fully submerged into the connector.02:02
wpwrakroh: have you made an uSD-card board too ? :)02:02
wolfsprauland then the only way to get them out it to push them further in, before they are 'kicked out'02:02
wolfspraulif you remove that mechanism, and you insert a regular microSD flash card, you may not get it out at all anymore :-)02:03
wpwrakyup. ah yes, and without the push-lock mechanism, you also don't need to be as precise02:03
wolfspraulbecause you have no way to grab the top and bottom side of the card02:03
wpwraka holder without that mechanism probably has an opening that lets you grab that raised border of the card :)02:03
rohwpwrak not yet. but i know the mechanics and its _small_02:03
wpwraklook at an uSD card. their thickness is not entirely uniform.02:04
wolfspraulI still don't totally get it.02:04
rohwpwrak i think they are intended to lock on that small nose on the side02:04
wpwrakroh: that's what we have CNC mills for :)02:04
wpwrakroh: yup02:04
wolfspraulI also know microsd connectors that open up on the top side, like in the freerunner for example02:04
wolfspraulthe standard says 1.0mm I think02:05
rohthe 'nonuniform' thickness is 'the traces of the pcb it consists of' and its thicker on the end, so you can grab it somewhere02:05
wolfspraulbut I'm sure there are tolerances, maybe by now multiple standards, etc.02:05
wpwrakwolfspraul: (open on top) they may be less extension-friendly. but it would have to be examined.02:05
wpwrakroh: i meant the latter02:05
wolfspraulyes I think so, much less extension friendly02:05
rohfor the 'push only' cheapo holders which dont 'push-lock' but 'push to the end and make it flush with some border' holders02:06
wolfspraulthat's why I was surprised you had some arguments against the push-in locking mechanism02:06
wpwrakroh: that's exactly what i'm looking for02:06
rohit think both mechanics exist. i like the 'more complicated' one more. gives a better 'warm and fuzzy feeling'02:06
wolfspraulI'm not sure what else you want, unless just no mechanism at all, and then I don't know how you can get a fully inserted card out.02:06
rohwpwrak i guess its a 'asian sourcing madness' for beginners job02:07
wpwrakadam to the rescue :)02:07
rohdont you have some intern for that? ;)02:07
wolfspraulroh: so you know some connectors that don't have this 'push in a little deeper to kick out' mechanism?02:07
rohwolfspraul ive seen it done.02:07
wolfspraulin that case, how do you get a regular, fully inserted, microSD card out?02:07
rohthe SDCARD to µSD thingies are that way02:08
rohwolfspraul fingernail02:08
wpwrakwolfspraul: basically all three types exist - top load, front load that clicks, and front load that just stays with friction. they all are designed to actually work :)02:08
rohits only in  'on the edge' use02:08
wolfspraulroh: ah OK. true! those little plastic adapters from micro to full-size.02:08
wpwrakgood example02:08
rohi guess those are just contacts on the board and the frame comes from the case parts02:08
wolfspraulthen they have an opening on the top02:08
wolfspraulbut of course you first pull out the entire full-size SD card in that example, so you get top access02:09
wolfspraulotherwise you couldn't get the microSD out anymore02:09
rohi still dont like that socket as internal extension conn.02:09
wolfspraul(not easily at least)02:09
rohnot really frickel-proof.02:09
rohcant you just make some nice layouted pads on the board?02:09
rohbigger for clumsy soldering. nice soldermask for avoiding tin bridges02:10
rohsome points with ground in distance, to mount a 'extension card' with some solder like the rfm12 are also intended for 'lay on board, put some tin around, done'02:11
wolfspraulroh: let me tell you something about the 'nice pads' that are seemingly always missing. the problem seems to be that you always just realize after the fact which 'nice pads' should have been put in which place on the board.02:11
rohnot meaning that one should use the rfm12 layout neccessarily. but the generic mechanics like a 'stamp'02:11
qi-bot[commit] Werner Almesberger: Be more careful with "volatile". It still worked, but was not a good example. http://qi-hw.com/p/ben-blinkenlights/979a63102:11
wolfspraulit seems to be very hard to predict in advance which pads will later turn out to be desperately missed, and where02:12
rohwolfspraul since it comes up so much, it seems to be something important02:12
roh*scnr* ;)02:12
wolfspraulsure sure, it's just my general feeling on this.02:12
wolfspraulit's the thing you always think about too late02:12
wpwraki think you can't solve the pad choice problem02:12
wolfspraulfor example we are working on Xue now, the camera02:12
wolfspraulwho is really thinking _THAT FAR_ ahead right now?02:13
rohthe neo had it right. spi, i2c, a irq, and you can do gpio on nearly all of them if needed.02:13
wpwrakbut converging on uSD as extension connector is something i like02:13
wolfspraulbut I guarantee you, in 1 year, 2 years. we will all be scratching our heads "why didn't we add this or that pad here or there"02:13
rohwhat was missing is power. and a spec how much you can draw02:13
wolfspraulyeah, why??? :-)02:13
wpwrakmake it a bit more flexible by merging an spi in, too. maybe i2c. so you can reconfigure for different boards.02:13
wolfspraulbecause it's very very hard to think that far into the future, although in hindsight that's still hard to accept02:13
rohwpwrak does microsd allow for i2c or an interrupt or is it just a glorified spi which cant speak spi anymore due to bad chipset design?02:14
wpwrakroh: uSD connector doesn't mean that you have to speak uSD on it02:14
rohdoesnt neccessary be a real i2c 'spec' .. can be something custom... dunno if there are free pins or so02:14
wolfspraulwpwrak: you and Ornotermes talked about bringing SPI and I2C to those pads as well. Does that work electrically? sounds like a totally good idea to me, if doable.02:15
wpwrakroh: just connect pins of the different engines together. e.g., MMC_CLK = SPI_SCLK = I2C_SCL02:15
rohwpwrak can you tristate unused ones?02:15
wpwrakwolfspraul: here's your answer :)02:15
wpwrakroh: sure02:15
wpwrakthe jz4720 is about as flexible as the 2442. no surprises there.02:16
rohwpwrak hm.. then its just a 'why that conn, and not some 0815-amp or molex conn?02:16
wpwrakin fact, it has a few nice extras. e.g., port in separated from port latch.02:16
wpwrakso if you do screw up your LED transistor, you don't need that certain kernel hack ... ;-)02:17
wpwrakroh: because you already have that connector02:17
rohi have? nope.. for nothing but flash02:17
wpwrakroh: add an internal one as well and you can a) choose where your board goes, and b) if you need an uSD, put it in the other02:18
rohand its mechanically not really nice. hard to 'make'02:18
rohtrue.. its nice if one can just push another flash in there02:18
wpwrakroh: i think we can figure that out :) and you can always make precision-milled adapter boards. most people will want to connect something clunky anyway, so you'll want a cable.02:19
rohhm. true.02:19
wpwraknot sure how FR4 compares to the uSD card's plastic for wear02:19
rohis there free pcb space?02:20
wpwrakin principle, it should be tougher02:20
wpwrakfor the internal connector ?02:20
rohso one could 'make a real connector solderable' ?02:20
rohsomething more mechanically stable than microsd? doesnt need to be populated02:20
wpwraki would make something a bit like the blinkenlights board. a bit longer. then solder some fpc on it. you can then solder the fpc to your board.02:21
wpwrakall very very simple. so you can produce a few k and just sell them by the bag.02:21
rohdoesnt even need to be useable with the regular case without cutting. i'd also add the bat-pins on there. so one could make a nanonote like the 'control terminal' with a plug connector into something 'more clunky'02:21
rohlike a rc-car or so.02:21
wpwrakfor battery, you could make a fake battery. block of wood with contacts.02:22
wpwrakattach it to your rc killer robot woth nails ;-)02:22
wpwrak(the block, not the ben :)02:22
bartbesand make it have the contacts for uart02:23
wpwraksoon we'll need a cpld for the multiplexing ;-)02:23
rohwpwrak something like that, but think of 'the sdio, the serial, 3pins of battery... all on one connector which one can solder in the battery holder place02:24
wpwrake.g., for ben-wpan, i'd want MMC_CLK, SCLK, MOSI and MISO together, nSS, and 2 GPIOs02:25
bartbesroh: like this: http://www.vibrant.com/images/cables/colorful-mess-zenji.jpg02:25
rohthen your mechanics only need to be very simple, and the connector could even be something like 2x7 pins in 2mm spacing02:25
wpwrakbartbes: precisely ;-))02:26
wpwrakbetter not to overengineer this. you could also have a path field nearby where you can rearrange the assignments by moving/placing 0Rs.02:26
rohi just got a device which got a 2x5pin 2mm header below. there is vbat, gnd, serial on cmos levels and usb on there. quite practical for debugging.02:26
wpwraknot extremely nice, but better than nothing if you need something not anticipated02:27
wpwrakplus, you could also run patch wires to the patch field and bring out completely new TPs02:27
rohaslong as its a exposed pad or resistor-end on the same side of the pcb, everything is already silver. the rest is gold and platin coverage for 'hack here' stickers02:27
wpwrakexpensive stickers if they need a gold and platin substrate ;-)02:29
rohhrhr.. i sometimes think my tax-guy should use those for the money he costs me02:30
wpwrakgrmbl. mail still down :-(02:31
rohanyhow.. just ideas. i'd love not to need hot-gluing those connectors into dremel-ed cases anymore02:31
wpwrakhehe :)02:31
rohjap. will talk with gismo once he is up to check with hetzner about chandra02:31
wpwrakkewl. thanks !02:32
wpwrakmaybe they moved power and air conditioning first, with the machines to follow in a couple of weeks :)02:33
wolfspraulroh: yes the qi server is also moving to a new datacenter, will be offline for 9 hours they say.02:33
rohi hope its some 'move' and not a dead disk or so02:33
wolfspraulbut they announce it very well 1 month in advance02:34
wpwrakdead disk would be ugly02:34
wolfspraulafter the move we will have ipv6 and then I can jump into enabling that...02:34
wpwrakwhee !02:34
wpwrakjust in time for ben-wpan :)02:34
kyakhi everyone02:35
wolfspraulcan you extend the range to ca. 20,000 km so we can reach the server from any ben-wpan in the whole world?02:35
wpwrakno problem. lemme do the math ...02:35
kyakxiangfu: yesterday i flashed the 'clean' image i built, mkfs* still doesn't work -\02:36
xiangfukyak: same here.02:36
kyakdo you have any ideas?02:36
kyakor any feedback from openwrt?02:37
wpwrakwolfspraul: about 5-10 MW output power should be enough :)02:37
wpwrak(with an omnidirectional antenna)02:38
kyakwpwrak: will it also make me a "vegetable" for the rest of my life?02:39
kyaki mean, being near this power source02:39
wpwrakif you like your vegetables really well cooked .. :)02:39
roh5-10 MW will make sure you will not have kids. or something left of the house the device tx-ed in02:40
wpwrakor need a pension plan for the radio operators :)02:41
rohwpwrak you won't even have to think about regulations. no need to worry!02:42
wpwrakhmm, i better get a nap. if everything goes well, then i'll have to be on standby for fedex in a few hours.02:42
xiangfukyak: I will send email to openwrt mailing list later. see if anyone have the same problem.02:42
xiangfu(search mailing list first, then send email  :)02:42
kyakxiangfu: ok, thanks!02:42
wpwrakroh: yeah. if they want to complain about anything, they should just come to the station ;-)02:43
wpwrakbtw, for connecting of bulky stuff: http://www.slashhome.se/p/uSD_breakout/02:44
wpwrakhe didn't use an fpc, but that the general concept looks right. also the photography is quite good :)02:45
wpwrak(note the cherenkov radiation around the acid)02:46
wolfspraulabsolutely beautiful pictures he took02:55
wolfspraulOrnotermes: congratulations for the nice pictures you have on your page! http://www.slashhome.se/p/uSD_breakout/04:47
OrnotermesThanks :)04:47
wolfspraulwhat license do you put these pictures under? do you mind if they are uploaded to the qi wiki (similar policy like wikipedia regarding pictures, i.e. cc-by ...)04:47
wolfspraulI couldn't find a license note on your page04:48
Ornotermesi have actually not read enough about cc licenses to decide yet04:48
wolfspraulok sure no problem, take your time04:49
wolfspraula good compromise between various extremes would be a simple CC-BY license04:50
wolfspraulfor inclusion in places like Wikipedia (or the Qi wiki) you should not use the non-commercial or no-derivatives restrictions, if you want your stuff to be included that is04:51
Ornotermesthe diffrence between cc by and cc by-sa is that cc by-sa requires derivated work to be released under the same license?04:57
wolfspraulOrnotermes: yes05:10
wolfspraulthe -by-sa is a copyleft license05:10
wolfspraulalthough I am not aware of any serious enforcement activities on the side of creative commons. and without that it might as well be wishful thinking.05:11
wolfspraulfor my own stuff, maybe I'll just use cc-by in the future, not cc-by-sa. I need to find out whether the -sa has any real meaning, and if it doesn't I'll remove it :-)05:12
Ornotermeswell, for most my nanonote related stuff i think cc by will do05:13
Ornotermesbut i will add it as an per image license05:14
wolfspraulgreat! with cc-by I think you make an excellent choice.05:15
Ornotermesbut it's a hard pick to do05:18
Ornotermeslicense for avt2 is cc by-sa05:18
wolfspraulOrnotermes: why hard? what do you prefer?05:45
Ornotermescc by gives mighty freedom for the next person, cc by-sa is a bit more restricted but the freedom reaches longer05:47
Ornotermesbut i guess the sa add on con be rather weak any way05:49
Ornotermesi think i license the text as cc-by and images as cc-by-sa(for now), how does that sound?05:57
wolfspraulhe :-) I never thought about it in that detail actually...05:58
wolfspraulyes it sounds good!05:59
Ornotermeswell, the text can any one reproduce, nothing special... but i don't want people to just slightly edit them to change the license06:00
Ornotermess/them/the images06:05
rafaOrnotermes: and you have a real pc keyboard.. like men should have ;)06:08
rafareal men06:08
Ornotermesrafa: i need some thing to defend my self with, you know ;P06:09
rafahaha.. yes.. i know.. i have a similar one06:09
rafaand it will survive for ages06:10
Ornotermestwo things survive nuclear wars, cockroaches and Model M keyboards. ;)06:11
wpwrakwolfspraul: that nobody is enforcing cc-by-sa today doesn't mean that nobody will tomorrow :) (see gpl-violations)08:00
Ornotermeshi wpwrak08:00
wpwrakwolfspraul: and then it's very useful to have old stuff that has found its way into many designs. not to screw the guys who comply 99% but get 1% wrong, but to prove in court 1% of those who happily violate the full 100%.08:01
Ornotermeswpwrak: http://gallery.slashhome.se/main.php?g2_itemId=291008:02
wpwrakwolfspraul: some of gpl-violations.org stories are just scary. e.g., that set-top box kinda thing in austria. they treated it like something proprietary. once forced to comply with the gpl, they - among other things - produced a document listing just what they had included. it read like a "who is who" is free software.08:03
wpwrakOrnotermes: heya ! that looks much nicer now ! :)08:04
Ornotermesyes, but i probably don't have time to test it today08:05
Ornotermesbut my old card fits i both my htc hero and htc tytn :)08:06
wpwrakwolfspraul: of course, the things on which legal action was taken/threatened was just the tiniest portion of the whole catalog. something that's absolutely bullet-proof so that they have no choice but to admit that they did something wrong. (well, some persist until they get a kick from the judge.)08:07
wolfspraulI doubt cc will ever put any serious enforcement action behind -sa, but let's see. I'm all for -sa.08:08
wpwrakOrnotermes: hehe, i smell the beginning of a multi-billion dollars industry. breakout boards for <brand>-<model> :)08:08
wolfspraulwpwrak: don't hold your hopes too high.08:08
wpwrakwolfspraul: cc don't need to do anything. they don't own the copyright anyway.08:08
Ornotermesone for all i much better :)08:08
wpwrakwolfspraul: again, see gpl-violations.org. it was harald who started fighting back, not the FSF.08:08
wolfspraulI have heard of several cases where handset manufacturers and/or operators were actively trying to _limit_ the possibilities of the SD port, SDIO in particular.08:09
wpwrakwolfspraul: and his ammunition was netfilter and later on also initrd and dosfs.08:09
wolfspraulthat's why a company like Spectec who is doing lots of SD cards will stay away from using SDIO, and instead use the simplest possible block transfers and proprietary drivers.08:09
wpwrakah, now on uSD ;-) okay, let's drop the licensing discussion :)08:10
wpwraknone of hese restrictions matte as long as you can bit-bang ;-)08:10
wolfspraulfor us it will work, I'm just explaining why in the bigger industry I think it cannot easily take off (like you said with "breakout boards for <brand>-<model>)08:11
wolfspraulmany times the lack of extendability is very much a feature08:11
wolfspraulI was disappointed to find out from Spectec, for example, that they were moving away from SDIO and towards their own (of course closed) drivers.08:12
wpwrakyeah. it's a control thing. not sure if it really matter in practical term. i mean, who but a small group of freaks would really have some peripherals stick out of their designer phone ? :)08:12
wolfspraulnot because they wanted to, but because the OS/kernel in the mobile device sometimes also has the function to 'manage' (=restrict) certain things08:12
wolfspraulwe just see wires, and signals. That's good :-)08:13
wpwrak(spectec) oh, that's sad08:13
wpwrakOrnotermes: nice. next step: go to 50 mil spacing and solder the ribbon cable directly to the board :)08:16
wpwraka bit like the long rounded pads on the board shown here: http://www.almesberger.net/misc/idbg/08:17
Ornotermesi was actually thinking about that08:17
wpwrak(the big ones on top are 100 mil, but the ones below are much narrower. a bit too narrow for ribbon cables in this case, though)08:18
wpwrakpackage still sitting at customs. mail still down. and the people who are on the road this morning seem to be very unhappy about the people who are on the road this morning, if all the honking is any indication :)08:19
Ornotermeswpwrak: in the future i was thinking of a tiny board in size of a uSD card vith a ffc soldered to it (to be able to fit in elecronics with sd ports under battery and such08:21
wpwrakyes, that would be perfect. it's also easier to solder :)08:22
Ornotermesbut thats a bit in the future08:22
wpwrakOrnotermes: sounds good. i should have an an actual usd-attached peripheral soonish. (ieee 802.15.4 prototype board) just waiting for components ...08:25
wpwrakwell, i can already start the design :)08:26
Ornotermesnice :)08:26
wolfspraulOrnotermes: what is a 'ffc'?08:30
Ornotermeswolfspraul: http://image.made-in-china.com/2f0j00pBzTlwWCCgcM/FFC-Cable.jpg08:30
wolfspraulinteresting - what is the difference between ffc and fpc?08:33
wolfspraulhmm, OK.08:39
wolfspraullearnt something :-)08:39
Ornotermeswpwrak: http://gallery.slashhome.se/main.php?g2_itemId=291908:53
wpwrakhehe :) now this is a size i like :)08:57
Ornotermesthink that will be the design i use08:57
wpwrakwhat's the spacing ? 50 mil ?08:58
Ornotermesif i done thigs right, yes :P08:58
wpwraklooks roughly okay. my led board also used a 50 mil spacing for the leds.09:00
wpwrakoh, but aren't your contacts upside-down ?09:01
wpwrakand there's one more thing to consider: a coating for the exposed traces. otherwise, they can short with the case09:02
Ornotermesby the way, did you see that uSD power was controlled by PD02 on CPU?09:02
wpwrakyeah. very very useful feature :)09:02
Ornotermesupside down?09:03
Ornotermesit shows the traces from topside09:03
wpwrakis copper on top or at the bottom ?09:04
wpwrakthen it's okay09:04
Ornotermesand i dont think there is anythin to short it with on ben09:04
wpwrakthe metal casing of the usd holder ?09:05
Ornotermesbut in production there should be a solder mask09:05
wpwrakah no, only on the other side09:05
Ornotermesand maybe a shielded leyer on top09:05
wpwrakyup. a bit of a ground plane would be nice.09:07
Ornotermesbut i think that would require a 0.7mm (or even 0.6mm) pcb09:08
Ornotermesi measure the uSD to be 0.76mm thich at the thin part09:09
wpwraknot sure an extra copper layer changes much09:12
wpwraklet's see how thick that is ...09:12
Ornotermesdont forget a mask on that too09:13
wpwrak36 mm for 1 oz copper09:14
wpwrakyou don't need a solder mask :) just a bit of tin. the more contact it makes, the better09:15
Ornotermesah, yes09:15
wpwrakerr .. 36 um09:15
wpwraklet's see how much the tin adds ...09:16
Ornotermesi was thinking of having it connected to to VSS09:17
wpwrakabout 10-20 um for tinning (manual)09:18
wpwrakyup. vss = shield.09:18
Action: wpwrak hates slow customs. it'll miss the deadline for delivery today :-(09:19
wpwrakany extra tin will wear off with time :) so about 870-880 um for the whole board. not too bad.09:20
Ornotermesit's still about 100 um thicker then a real world uSD card09:24
Ornotermesbut if it still fits its really good09:25
wpwrakit's a feature - extra contact pressure for improved reliability :-)09:26
Ornotermesi see :P09:32
Ornotermesoh well, i have too leave. bye09:38
wpwrakah, package finally got out of customs. let's see if they'll try to deliver it today.09:43
wpwrakand mail is back, too. the day is looking up :-)09:48
wolfspraula lot of stuff released today :-) I got a Milkymist One released in China, a ML-401 was released in Sweden. Now your package :-)09:48
wolfspraulI have one more nasty package stuck in Shanghai, days of struggling with those customs retards. Maybe released tomorrow, we see...09:49
wpwrakbatteries ? :-)09:49
wolfspraulno just a NanoNote09:50
wpwrakwow. 231 new mails, 0 of them on qi-hw.09:50
wpwraklet's change that09:50
wolfspraulwhat do you mean with "on qi-hw"?09:53
kristianpaulwpwrak: hey thanks for sharing results, so i will let my router as it is now :)09:54
wpwrakon fedex vehicle. yeah !09:57
wpwraknow, a quick morning shower :)09:57
wpwrakah, mails are just slowly arriving. that's why there seemed to be none in qihw.10:35
wpwrakxiangfu: welcome to u-boot hell :)10:38
xiangfuwpwrak: :)10:38
wpwrakit's kinda ironic that qi-hw doesn't use something like qi10:39
xiangfuwpwrak: some of the u-boot code are old. like the mmc struct. it's spend me sometime to find out :)10:39
wpwrakxiangfu: i had my share of fun with u-boot in openmoko. it's amzing how much time you can spend on solving problems in u-boot ...10:40
xiangfu:)  time to sleep. see you .10:42
wpwraksweet dreams ! :)10:42
kristianpaulah, wait a min do i need edit uboot in order to able linux acess gpio wich is not currently used?10:47
kristianpauli guess that is already mapped10:50
kristianpauldamm i need readm more..10:50
wpwrakfrom user space ? naw, you can do pretty much whatever you please. you may the register file and there you are10:50
wpwraki updated bbl.c to make things a little clearer10:50
kristianpaulwpwrak: yeah user space10:52
kristianpaulafter bbl update i see comments ! :)10:55
kristianpaulwpwrak: you pointed a cypress cpld the other day, can you remenber me the ref?10:59
wpwrakit's mcu with analog and cpld inside. psoc3 and psoc5.11:00
kristianpaulpaso :)11:01
wpwrakbut they're still not in mass production. just for evaluation11:01
wpwrakpsoc3 is 8051. psoc5 is arm (cortex)11:01
kristianpauli may start a NO-ARM Campaing in this channel :)11:02
wolfspraulit's called Milkymist11:02
Action: wpwrak celebrates the arrival of the toys :)11:29
emebwhat kind of toys?11:33
wpwrakemeb: the kind yuo buy at digi-key11:34
emebthose are good.11:34
emebparts for new circuits?11:34
wpwrakemeb: yup. lots of rf stuff. also lab supplies. adapters, antennas, and such.11:53
kyakok, i just built with uClibc- (instead of uClibc-0.9.32) and now things works again (mkfs*, specifically)12:07
kyakseems that upgrade to uClibc-0.9.32 was not smooth after all12:08
kyakuClibc-0.9.32 introduced support for nptl, could this be the problem?12:09
kyakwow! NanoMap runs on my laptop12:50
kyakneil is amazing12:51
wpwrakDocScrutinizer: question: in figure 5-1 at page 8 of http://www.atmel.com/dyn/resources/prod_documents/doc5131.pdf12:59
wpwrakDocScrutinizer: which of the two GND connections of the balun B1 would you call "DC feed" ?13:00
wpwrakDocScrutinizer: here's the data sheet of the balun: http://katalog.we-online.de/kataloge/eisos/media/pdf/748421245.pdf13:00
Ornotermesi'm back13:08
kristianpaulhey good that arrives wpwrak , so it takes 3days ?13:12
kristianpaulbest wishes in your coming hacking rf adventures :)13:13
gnutoohi kristianpaul13:13
wpwrakkristianpaul: ~30 h digi-key -> buenos aires, ~32 h customs, ~3 h delivery13:14
kristianpauli missed thet time13:15
wpwrakkristianpaul: here are the details :) http://www.fedex.com/Tracking?action=track&language=english&cntry_code=us&tracknumbers=45739681153013:15
kristianpaulgnutoo: wait tuxbrain he will apear eventually13:15
wpwrakkristianpaul: (best wishes) thanks ! :)13:15
gnutooah ok he's not there13:15
gnutookristianpaul, should I wait a long time, did they usually come here often?13:17
kristianpaulgnutoo: he lives here :)13:18
gnutooah ok13:18
kristianpauli think he is kind of busy now because some workshops13:18
gnutoowhich one from tuxbrain is he?13:18
gnutoobecause at fosdem I met 2 people13:18
gnutoois there some pictures13:18
kristianpauldavid pictures?13:19
kristianpaulgnutoo: http://blogs.fsfe.org/fellowship-interviews/files/2010/05/david-samblas.jpg13:22
gnutoook thanks13:25
kristianpaulgnutoo: are you planning buy a nanonote btw??13:26
gnutooI've too much devices13:27
gnutooand more incomming13:27
gnutoothey sell geekphone13:28
kristianpaulyour are here, thats good13:28
gnutooand I was asked to contact them13:28
gnutoofor SHR port13:28
gnutooor rather for evaluation of it13:28
kristianpauli see13:28
gnutoobecause he knows the device13:29
gnutooit will be faster if he pointes me to the right sources etc...13:29
DocScrutinizer( DocScrutinizer: which of the two GND connections of the balun B1 would you call "DC feed" ?) The sym GND is the "digital" GND which may usually carry DC components from both chip outputs - it shall be connected to same GND plane as chip. The asym GND is the RF GND and should treated similar to a dipole antenna - means this GND plane should be the outermost, possibly even case shielding (my educated guess)13:43
DocScrutinizerk, have to run13:43
DocScrutinizerwpwrak: ^^^13:44
gnutooDocScrutinizer, hi you're everywhere13:46
wpwrakhmm. all my gnds are the same there13:46
wpwrakwhat puzzles me is that they call one "DC feed". but i guess this would be the anymmetric one then. e.g., for phantom voltage to the antenna, no ?13:48
wpwrak(i just want to get the labels right. don't have any such fancy things in my system)13:49
kristianpaulwpwrak: do you find your tiny antenna finally?13:54
wpwrakah yes, i have two antennas now. one sma but only 2.4 GHz, the other 2.4/5 GHz but rp-sma. i also have a collection of gender benders for the latter kind of problem :)13:56
qi-bot[commit] Juan64Bits: PSU Modifying. http://qi-hw.com/p/xue/9161ab113:57
DocScrutinizerwpwrak: No the sym is the DC feed, as it would connect to VDD for open collector NPN driver14:03
wpwrakaaah, i see14:05
wpwrakthanks a bunch ! one more mystery solved :)14:05
DocScrutinizerbtw the balun 'datasheet' is a parody14:06
DocScrutinizersorry no idea what to do with that, might be useful for fab, not for EE14:07
wpwrakyeah. none of the balun data sheets i've come across make much sense :-(14:08
DocScrutinizertook me 30s and 2 glances to accept I can't even find real electrical pinout14:11
wpwrakDocScrutinizer: felt the same about it - that's why i asked you :)14:17
Action: wpwrak hates brown-out. especially while doing layouts :-(18:45
mthwpwrak: would a UPS help to bridge the time the power is out?18:46
wpwrakinstalling a better power supply already would18:48
wpwrakmth: i don't quite trust ups. they often introduce their own failure modes.18:48
mththey do, but maybe less frequent than the brown-outs18:50
wpwrakmy file server has an uptime of 45 days. it only notices the really bad ones. my workstation, on the other hand, never misses even the shortest one :-(18:51
wpwrakbut yes, i have a better power supply already waiting. just have to install it.18:51
wpwrakhmm. two of the screens didn't come up. guess it wants another reset ...18:53
Ornotermeswpwrak: i have etched my new layout now18:54
mthI think he's rebooting right now18:55
Ornotermeswpwrak: http://gallery.slashhome.se/main.php?g2_itemId=294318:56
wpwraknice :)18:56
wpwrakOrnotermes: i like the style of your photos. very good composition and lighting.18:57
Ornotermesthanks :)18:58
Ornotermesthat last card git a bit smudged so i had fix it with needle and knife18:59
Ornotermesbut i think my acid worked better today19:00
wpwrakthey can be moody at times19:12
Ornotermesi think it was a bit cold yesterday19:12
Ornotermesi have it in a big glass jar and put that in a plastic bowl and pour hot water in the bowl19:14
wpwrakOrnotermes: what acid are you using ? HCl+O2H2 ?20:02
wpwrakH2O2 even20:02
Ornotermesi think the english name would be sodiumperoxidesulfate20:07
wpwrakah, i think i heard of that one. not very common, though.20:10
Ornotermesthe most common around here20:13
wpwrakah, interesting. here, they still use ferric chloride. i was very happy when i discovered hcl+h2o2 ;-)20:17
uncloudedisn't that the one you can rejuvinate so you never throw it away?20:19
Ornotermeswpwrak: http://gallery.slashhome.se/main.php?g2_itemId=294820:23
wpwrakunclouded: supposedly yes, but i'm not so sure about that process. i find that it decays fairly quickly if i don't do anything special.20:28
wpwrakunclouded: it's so cheap and easy to make that i also don't really care :)20:28
wpwrakOrnotermes: looking more and more professional :)20:29
Ornotermeswho knows where it ends ;)20:29
Ornotermesit's not that big of a step to order a panel of those, just money :P20:30
uncloudedwpwrak: got to try HCl+H2O2 then.  I'm using Ammonium Persulphate at the moment20:31
wpwrakOrnotermes: with an fpc, you can solder everything on a pcb - also the header. then cover it with a bit of silicone or PUR and it'll look quite professional20:37
wpwrakunclouded: i tried ammonium persulphate once or twice but found it a bit slow20:39
Ornotermesyes, that could work, but how is the mechanical stability?20:39
uncloudedmaybe it is slow.  I thought I'd just made it weak20:41
wpwrakOrnotermes: for FPC and header probably better than what you have now. ah, i mean the part that goes into the prototyping board.20:42
wpwrakOrnotermes: so  uSD->FPC ... FPC->PCB->header20:42
wpwraksolder the header flat on the pcb, smt-style. that won't come off anymore.20:43
Ornotermesyeah, could be a good idea, from point of manufacture any way :)20:44
wpwraknot my day today. pc went crazy :-(20:51
kristianpaullinux crashed?20:51
wpwrakx server, maybe in cooperation with pcbnew.20:53
kristianpaulahh but thats not bad my x crashed once at month i think20:53
wpwraksomething's very wrong in my setup. since the last ubuntu upgrade, i have as many x crashes in a week as i'd expect to have in ten years ...20:54
kristianpaulmove to debian to :^)20:55
wpwraki think it's because i once manually added drivers. seems that the upgrade doesn't quite flush them out20:55
kristianpaulbbl i need reboot and test a xen image20:56
kristianpaulgood now i have X running with xen thanks to debian linux-xen maitainers :)20:59
kristianpaulhttp://blog.ninapaley.com/2010/08/31/four-freedoms-of-free-culture/ :)21:53
wpwrak git push22:53
wpwrakPHP Fatal error:  Allowed memory size of 262144 bytes exhausted (tried to allocate 523800 bytes) in Unknown on line 022:53
wpwrakduh ?22:53
wolfspraulthis just happened now?22:55
wpwraklemme try again22:55
qi-bot[commit] Werner Almesberger: Added components (transistor and balun) for the uSD variant. http://qi-hw.com/p/ben-wpan/7c1383d22:55
qi-bot[commit] Werner Almesberger: atusd/: first try at a uSD-based board. Improved RF design. http://qi-hw.com/p/ben-wpan/dd999eb22:55
qi-bot[commit] Werner Almesberger: atusd/cam/: CAM data for cutting the PCB. http://qi-hw.com/p/ben-wpan/eb6f06622:55
wpwrakah, this time it worked22:55
wolfspraulsorry might have been me22:56
wolfspraulI was a bit confused about PHP comment syntax, used the wrong thing in an .ini file22:56
wpwrakhah, gotcha ! ;-)22:57
wolfspraulwpwrak: well I'm not sure it was the cause, but since I was _just_ editing that .ini file when you reported the problem, and then the problem went away, maybe it was23:09
wolfspraullet's just keep an eye on it23:09
wolfspraulif your commits exhause memory somewhere, I will tweak more config options23:10
wpwrakah yes, editing and running often don't go together :)23:14
--- Fri Sep 3 201000:00

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