#qi-hardware IRC log for Friday, 2012-05-25

kyakreading about TI launchpad.. it has "touch-sense enabled I/Os" - wtf is that?09:27
kyakwell, MSP430G* has it.. i'm confused by "touch sense"09:28
wpwrakkyak: it probably means that someone figured out a way to connect a capacitative sensor09:30
wpwrakkyak: the leanest solution i've seen so far comes from atmel: just the internal ADC and a series resistor. and i suspect you don't need that resistor. so it's not all *that* demanding in the hw side :)09:32
kyakwpwrak: ok, it's clear, thanks! so probably "touch sense I/O" is just a marketing word :)09:33
kyaki could resist ordering it.. damn, it's so cheap09:34
kyakshipping included09:34
wpwrakthe real cost is in the development effort :)09:35
wpwrakat least, with a ben and ubb, you can easily make your own programmer (if they document the in-circuit programming protocol)09:36
kyakwell, i could run mspdebug on Ben09:36
kyakand program launchpad easily without knowing the programming protocol, right?09:37
wpwrakwhatever works, as long as it works :)09:37
cladamwwpwrak, http://dpaste.com/752070/09:47
cladamwwhen I searched 'package drawing' and 'land pattern' for Analog Device, Wolfson, Maxim, SiLabs about QFP48, i got some info so that I recorded INFO there.09:49
wpwrakyou should also mention the conclusion. e.g., if they agree09:50
wpwrakat the end of the day, a lot of those footprints correspond to some JEDEC standard. but there's a crazily large number of variants09:51
cladamwMAXIM has great completed lists about all package drawings and land patterns, AD has few but not completed. Micrel/Wolfson even has very poor on collecting land pattern.09:51
wpwrakwolfson also have some rather crazy variants ;-)09:51
wpwrake.g., page 13 of http://www.wolfsonmicro.com/documents/uploads/misc/en/WAN0118.pdf09:54
cladamwto finish m1r4's total three QFP48 parts(WM9707, KS8001, ADV7125), i did almost 99.5% done roughly but i merged your qfp.fpd -->09:55
cladamwhttp://downloads.qi-hardware.com/people/adam/m1/tmp/kicad/qfp.fpd09:55
cladamwi now very feel uncomfortable to wolfson, even AD, i very like Maxim's data base. so we could change a little dimensions based on datasheet's package drawings, if chip vendor has less info on land pattern, then i recommend to refer to Maxim, then to meet footprint reasonably.09:58
cladamwi found our m1r4's WM9707 and KS8001 both has the same package drawing.(almost :-) ) 09:59
wpwrakmaxim's pas length seems to be a bit excessively long, while ad's seems a bit on the sort side. (maxim+AD)/2 would seem about right ;-))10:00
wpwraks/sort/short10:00
wpwraks/ pas / pad 10:00
cladamwyeah ... 10:01
cladamwwpwrak, my qfp.fpd includes your SiLabs C8051F32x, but named "QFP$pins-$variant", do you think this naming is okay to you ?10:03
wpwrakthe maxim is in truth a JEDEC MS-026 BBA or BBC10:03
cladamwBBA for 32pins, BBC for 48pins.10:05
cladamwyes, i think maxim did a great work on them(almost the whole package, else chip vendors may not)10:05
wpwrakyes, sounds good. without a proper reference, i'd consider the silabs thing a pseudo-vendor standard anyway10:05
wpwrakBBA/BBC ah, right. maybe we can even get rid of the silabs critter then ... but lemme first check the 48 pin packages10:06
wpwraki find NXP very good. they have a weird naming scheme (they have a "SOT-xxxx" code for each package, which is not the same as what the rest of the world calls "SOT", but they adhere quite strictly to standards)10:07
cladamwif i didn't use your this Fped to edit correct land pattern, i'll never known how else famous chip vendors they are poor in such info collection referred to JEDEC or IPC. :-)10:08
wpwrak(NXP) even better, they often specify reflow and wave soldering footprints. than often explains apparently crazy variations in other vendor's footprints - they may use some ancient wave soldering footprint10:08
wpwrak;-)))10:09
cladamwaha ... NXP, I've not digged into it now. since I started to build module one-by-one from wiki m1r4 bom. so maybe later I'll definitely run into NXP. :-)10:09
wpwrakNXP also have some helpful background information. e.g., i found their design rules for QFN quite helpful, which include the solder paste islands and such10:10
cladamwso if you don't mind, I'll commit my qfp.fpd, but you may need to update your project used qfp.fpd. 10:10
wpwrakalright10:11
cladamwmmm ....  a more solder paste info is useful, nice news. I'll see it.10:11
cladamwman~ hehe ... i found i spent 2 days to work on this (qfp) but seems worthy and also learnt a lot that how your table v.s. vector works. :-)10:12
wpwrakQFP48 of maxim and AD have an identical package, only maxim have +/- 0.1 mm while AD have +/- 0.2 mm for D/E/D1/E1. L is the same.10:13
wpwrakso we should have only one footprint for maxim and AD10:13
wpwrakthat is, there should be an MS-026-BBC footprint that applies to both10:14
cladamwAD's (ST-48) land pattern is horrible, must be no one their internal to double confirm.10:14
wpwrakmy it was an "intern" who made it ;-)10:14
wpwraks/my/maybe/10:14
wpwrakyeah, the AD pattern looks like trouble. at least the chip would just fit if you have the minimum values for D/E ;-)10:19
qi-bot[commit] Adam Wang: qfp.fpd: added Maxim C48 and Analog Device ST-48 variant (master) http://qi-hw.com/p/kicad-libs/bd5c1c310:20
qi-bot[commit] Adam Wang: Merge branch 'master' of projects.qi-hardware.com:kicad-libs (master) http://qi-hw.com/p/kicad-libs/61fecbd10:20
wpwraki'd get rid of the AD pattern. there's no point in keeping that footprint if it's obviously wrong10:21
cladamwwpwrak, i think later we fix AD variant, meanwhile I'll go for another, if you see any err, let me know.10:21
cladamwaha ... okay10:21
cladamwwait ...10:21
wpwraki would just rename the Maxim_C48 to something like MS025BBC. or maybe just call it "QFP48" or "LQFP48" without variant, since this seems to be the "standard" size anyway10:23
wpwrakyou can then put references to suitable chips / vendor packages and footprints in INFO10:23
wolfspraulI printed out a set of m1 schematics on paper for medidation10:24
wpwrakhehe :)10:24
wpwrakpaper is great for reviewing. you see a log more than on the screen10:25
wolfspraulwill put it under my pillow10:25
wolfspraulsee whether that helps :-)10:25
wolfspraulbut anyway, first thing that struck me again is lack of page numbers10:25
wolfspraulso I had to number them by hand10:25
wpwrakinteresting approach for information acquisition :)10:25
wolfspraulthe font is a little on the small side, but just readable10:26
wolfspraulunlike say the m1-gps-extension :-)10:26
wpwrakyeah, we should label the pages. maybe not with numbers but with short names10:26
wolfspraulbut a bit bigger would be nice10:26
wpwrak(readable) that's A3 sheets for you :)10:26
wolfspraulbut overall it's really nice already, given that many schematics don't print at all or with all sorts of issues10:26
wolfspraulnah it fits nicely on A4, all fine10:27
wolfsprauland there is space for the font to be bigger10:27
wolfspraulbut the #1 issue is lack of page numbers, lack of license or any form of attribution10:27
wpwrakhmm. if you auto-scale the fonts, a lot of things will start to look weird10:27
wpwrakthe fonts have some readability issues, though. they could look better at exactly the same size10:28
wolfspraul#1 issue is page numbers, license, attribution10:28
wolfspraulshould we just write something manually on each page?10:28
wpwrakperhaps some bold individual could write a converter that takes each text field, calculates the exact size with the original font, then places a PS font scaled to just that size (aspect rate not necessarily 1:1) instead10:29
wolfspraulI also printed out the (quite simple) atben page, that one looks even better, I guess the fonts come out a bit bigger10:29
wolfspraulbut the atben one is also missing any form of license, copyright or attribution10:29
wpwraki use A4 sheets, so the font has the "natural" size10:30
wolfspraulthat's sad10:30
wolfspraulthe creators should be honored :-)10:30
wpwrakthe author is in the little text box in the corner, if you have that one (schhist suppresses it to avoid cluttering things even more than they already are)10:31
wolfspraulyes, I know10:31
wolfspraulbut that box is ugly and overloaded and we disable it for a reason in schhist10:31
wpwrakyeah :)10:32
wolfspraulbut maybe we should write *something* manually instead?10:32
wolfspraulotherwise it's just totally blank, which is also not right10:32
wpwrakdunno. perhaps in the case of M1, since it has a big overview sheet already10:32
wolfspraulI wouldn't mind if schhist picked up a textual change, that would also be the case already now for notes10:32
wolfsprauloverview yes, but again10:32
wpwrakfor smaller projects, it would just add more clutter10:32
wolfspraulabsolutely nothing about license, creators, copyright, attribution10:32
wolfspraulit's like something that fell from the sky10:33
wolfspraulone question about atben10:33
wolfspraulthe RFP and RFN wires are swapped10:33
wpwrakthe default is "no rights, no copying" so you have to look in the project (which you'll have to do if you're going to do anything interesting with the schematics anyway)10:33
wolfspraulis there a technical reason for that, or just a random effect of the drawing?10:33
wolfspraulmaybe turning B1 upside down would be even uglier? don't know10:34
wolfspraulmaybe you could just turn the right side of B1 upside down?10:35
wpwraki did it this way to help with the layout10:35
wolfspraulthose pin numbers don't look like on the device anyway10:35
wolfspraullayout?10:35
wpwrakthe crossing. but yes, i could have flipped B1 and then uncrossed RFP/RFN 10:36
wolfspraulwhat does the /RF in C1/C2 stand for?10:37
wolfspraulI mean what parameters are we watching for in those caps?10:37
qi-bot[commit] Adam Wang: qfp.fpd: get rid of the AD land patter and rename variant to name (master) http://qi-hw.com/p/kicad-libs/d4be1a410:38
wpwrakmy schematics have the balun drawn like in atmel's reference design. (but atmel don't cross RFP/RFN)10:38
wolfspraulthe R1 is 0R, and C12 is NC - is this left purely for tweaking & playing reasons?10:38
wpwrakC1/C2 good question ;-) the parameters we're "watching" for are somewhere three to four orders of magnitude beyond the parameters whet component makers actually specify10:38
wolfspraulon the RF path, I think at least I would feel better if there would be a note like "can safely be removed" if that is so10:38
wpwraks/whet/the10:38
wolfspraulotherwise one might thing that whatever effect is meant to be achieved with those 0R or even the NC pads :-)10:39
wolfspraulmight think10:39
wpwrakhaving a place for inductors and caps at the antenna side is a common arrangement for matching10:39
wpwrakand yes, if you make your own derivative design, you're free to omit them :)10:40
wolfspraulduring design maybe, ok. but I think a note "can be removed" would clarify things, no?10:40
wolfspraulsince this is on the RF path10:40
wpwrakthere are always implementation details like this that creep into schematics. schematics aren't expositions of pure theory :)10:40
qi-bot[commit] Adam Wang: INFO: removed AD QFP info (master) http://qi-hw.com/p/kicad-libs/6c2655c10:40
wolfspraulotherwise I can lookup 20 effects at least that one 'might' want to achieve with a 0R resistor, or the empty pads10:40
wolfspraulmy question is: should we add a note "can be removed"?10:41
wpwraknaw, i think it's just a question of recognizing that pattern. it's common in RF design10:41
wolfspraulI think that may be helpful, if that is what was last in the designers mind about R1 and C1210:41
wolfspraulsure10:42
wolfspraulI think you don't understand my point :-)10:42
wpwrakyou'll even find some that have something like four caps and such there. atben is very light on obscure RF critters :)10:42
wolfspraulnot 'obscure'10:42
wolfspraulyou design it, you have some state of your thinking about the design10:42
wolfspraulthat is either conveyed through the schematics or not10:42
wolfspraulit seems that right now you are not aware of any effect of R1 or (the NC) C1210:42
wpwrakyes, but that's basic RF design knowledge10:42
wpwrakif you don't know what these parts are not essential, then you probably shouldn't remove them anyway, because your circuit may need tuning ;-)10:43
wolfspraulalrigh then, I tried :-)10:43
wolfspraulso the /RF then?10:44
wolfspraulit stands for?10:44
wpwraki know their effect (in general terms). it's for adding small inductances or small capacitances to adjust the properties of that signal path. there are incidental parameters we don't know. so if one goes off the scale, we may need too tweak things10:45
wolfspraulsure but your design is stable and works10:45
wpwrakthe /RF means that the caps are specifically designated for RF applications10:45
wpwrakunfortunately, that's about as good as it gets, in terms of specifics10:45
wolfspraulyou could put even more matching networks there and leave them all empty, but what would be the point?10:46
wolfspraulby leaving R1 and C12 there you want to remind people that the matching is not done properly?10:46
wolfspraul:-)10:46
wpwraksure. i have a basic matching network there. if you have other means to accomplish the same or if you're certain you don't need one, then drop it by all means10:47
wolfspraulhow do you source the caps then, with /RF in digikey?10:47
wolfspraulyes [matching]10:47
wolfspraulbut it seems that R1 and C12 are not needed10:47
wolfspraulso leaving them there may add confusion10:48
wolfspraulsomeone with enough RF experience will easily know how and where to add matching networks anyway10:48
wolfspraulbut leaving those seemingly unneeded R1 and C12 there conveys a strange message, at least to me10:48
wpwrakR1 and C12 remind people that RF is tricky. e.g., if you add some case, you may find that the device is detuned. then you need to make adjustments.10:48
wolfspraulhe10:49
wolfspraulthat I fully understand10:49
wpwrakof course, if you're able to make meaningful adjustments, it's rather likely that you understand what R1 and C12 mean :)10:49
wolfspraulbut trying to get that across by leaving two unused parts there is strange :-)10:49
wolfspraulwell that's the thing, they mean nothing10:49
wolfspraulI guess10:49
wpwrakit's pretty much standard practice10:49
wolfspraulfrom your answer10:49
wolfspraulstandard practice is to leave one empty resistor and one empty/nc cap as a reminder for a matching network?10:50
wolfspraulI will pay attention to that10:50
wpwrakthey're not only reminders. they have real-life footprints.10:50
wpwraki need these components in the schematics for kicad to know what to do with the footprints.10:51
wolfspraulyou think someone may want to put a part there if they add a case?10:51
wolfsprauland that's standard on an RF path to leave some slots for matching parts?10:52
wolfsprauldoesn't sound very credible to me, but I will keep an eye on it10:52
wolfspraulyou wouldn't be able to predict what parts are needed, and in fact the empty pads themselves may be the source of uncertainties10:53
wpwrak(case) yes, or if they want to optimize the circuit further. e.g., there may be RF leakage issues i'm not aware of. my lab isn't set up to do full FCC compliance testing.10:53
wolfspraulsure10:53
wolfspraulthat I understand10:53
wpwrakhmm, i had typed "complicance". thank you, Freud :)10:53
wolfspraulbut you cannot predict whether that one resistor and one cap is what you will need10:53
wolfspraulit looks a little helpless, unless this is really a common way to do it that I'm not aware of10:54
wolfspraulbut I'm learning, and now I will pay attention to this10:54
wpwrakand yes, leaving slots is common, particularly in designs that may experience environmental variations or experiments10:54
cladamwwpwrak, do you have SOIC module ? i saw SSOP but seems no SOIC. if you have it, point me the link. tks. :-)10:54
wolfspraulback to the /RF cap - so how do you source it then?10:54
wolfspraulI should look into the sources, sorry haven't done that yet10:55
wpwrakif you need a more complex matching network, you'll need a redesign anyway. but L and C (the R is really a placeholder for an L) are your starting point10:55
wolfsprauljust noticed the /RF and didn't know what it means10:55
wolfspraulyes, I know10:55
wolfspraulit seems half-baked to me10:55
wpwrakthe /RF selects a part that has the X=RF field set. "X" is a field for "additional characteristics". so when building my characteristics database, i set that field for caps from an "RF" series10:56
wolfspraulhmm10:56
wolfspraulI see10:56
wolfsprauljust from that series10:56
wolfspraulwithout going into the specifics of the parameters10:56
wpwrakthe quantitative parameters don't really help10:57
wpwrakthey tell you how the part behaves at, say, 1 MHz. we're running it at 2.4 GHz.10:57
wpwrakbut the vendor stating that it's for HF, microwave, RF, whatever, tells you that you're less likely to experience nasty surprises with that part than with one that's not rated for such uses10:58
wpwraknasty surprises may including one batch working just fine and the next not10:58
wpwrakand yes, it looks very half-baked to me too :)10:58
cladamwwpwrak, if no SOIC module, i'll build it same likely SSOP. :-)10:59
wpwrakcladamw: (SOIC) lemme see if i have any of those fat pigs somewhere ...10:59
wpwraknope, don't think so11:00
cladamwwpwrak, yeah ... i didn't find it after searching wernermisc and ben-wpan. tks. :-)11:01
wpwrakwolfspraul: (RF caps) here's a typical collection of such critters: http://www.johansontechnology.com/images/stories/catalog/JTI_CAT_2012_MLCC_HighQ.pdf11:12
wpwraknote the "DIELECTRIC" table on page 1. "ultra high Q", "high Q", ...11:12
wpwrakand then you go to page 10, to learn that all NP0 have Q >1,000 @ 1 MHz, Typical 10,00011:14
wpwrakand that's all you get. you just have to trust them that they'll also behave in multi-GHz circuits :)11:15
wpwraksome other descriptions read "ultra-high Q", "mid-high Q", and "excellent high Q". some engineering poetry, i guess :)11:33
wpwrakinteresting .. the silabs LQFP32 is 0.1 mm smaller than the maxim (JEDEC 95 MS-026 BBA) part11:51
wpwrakyet silabs claim their is MS-026 BBA, too :)11:52
wpwraks/their/theirs11:52
wpwrakah, no. they're the same. confused the column of typical with maximal sizes.11:53
wpwrakheh. we could express all this in terms of the IPC-7351 formulas :)12:12
kristianpaulthey are tiny but i still can read on my screen13:10
kristianpaulbtw i'll add license, already agreed one with Artyom, but just a license.txt file or also add in the sheet?13:11
wolfspraulstart with license.txt that's great13:15
wolfspraullook how long it took you to even find out what it is :-)13:15
kristianpaulyou like i separate the "blocks" in sheets?13:16
kristianpaulanyway after the cutoff just the half will remain..13:16
kristianpaulwolfspraul: the license you mean?13:18
kristianpaulor the font?13:18
kristianpaulerghh i cant search in those pdf..13:20
wolfspraullicense13:21
wolfspraulobviously knowing the license is by far more important than size of font, no? :-)13:21
wolfspraulso one by one. seems the license and copyright info for m1-gps-extension is clear now? great!13:22
kristianpaulyes, share a like by license is13:22
wolfspraulwhat I was talking about in the m1 and ben-wpan was at another level, because as wpwrak already said the license and copyright is clear there, just the pdf is a little unspecific on those things13:22
kristianpauli se13:22
kristianpaule13:22
kristianpaulbetter write license as a schematic component, so that way is kept13:30
wpwrak(license as component) that's an interesting idea :) lemme see if you can actually make it such that it doesn't end up on the BOM by default14:23
wpwrakyeah. make its component reference start with # and set the field (reference) to invisible. that should do the trick.14:28
qi-bot[commit] Jiva Nath Bagale: fixed guile2 Makefile to use guile from toolchain instead of host machine (master) http://qi-hw.com/p/openwrt-packages/e88e68e16:13
kristianpaulkyak: adc from launchpad plus a nice plotting app in the nanonote (i guess we already have)16:29
kristianpaulmsp430 have adc/dac i read somwhere..16:30
kristianpaulkyak: looks interesting idea for uses in the field..16:30
kristianpauls/field/wild?16:30
kyakkristianpaul: there already a python+gnuplot plotting script here: http://losinggeneration.homelinux.org/2010/07/02/msp430-launchpad-on-linux/. Both python and gnuplot are available on Ben, so it should work :) 16:42
kristianpaulnice16:42
kyakit seems that the USB connection is not very reliable.. You have to replug every time, as mentioned here: http://dbindner.freeshell.org/msp430/index.html#Talking_to_MSP43016:43
kyakalso mspdebug only work via USB as i understand16:43
losinggenerationkyak: I can vouch for it being less than reliable :)16:43
kristianpaulit tought over uart as well.. shame16:44
kyakso probably there is a need for another software programmer, as werner mentioned16:44
kyaklosinggeneration: do you have an idea what might be the problem?16:45
losinggenerationoff the top of my head: no. It's actually been quite some time since I've last picked up the launchpad, so I can't recall the exact issue besides it not working well without unplugging it and plugging it back in16:47
kyakit will lead to loose usb connector and unusable board at the end16:49
losinggenerationIIRC it was acting as if  the connection to the device was blocked16:49
kristianpaulpehaps is because uart is emulated in software?16:51
kristianpaulperhaps*16:51
kristianpaulfirst time i pluged the thing, my ubuntu freezed for about 10 seconds :-|16:52
kristianpaulperhaps is a cdc running in other TI MCU.?16:52
qi-bot[commit] Werner Almesberger: bacon/prog/: simple PIC 18F{2,4}xJxx programmer (for the Ben Nanonote) (master) http://qi-hw.com/p/wernermisc/6bdfea616:56
kristianpaulah no16:56
wpwrakkyak: this may be useful as an example: http://projects.qi-hardware.com/index.php/p/wernermisc/source/tree/master/bacon/prog/16:56
kristianpaulit have its own Serial-to-USB chip16:57
rjeffriesThis program sounds interesting. Will it work on a Ben NN? http://www.kiwix.org/index.php/Main_Page17:05
kristianpaulthere was a reader for kiwix on the ben17:08
kristianpaulactually need some loving..17:08
kristianpaulbut keep in mind the interesting and long process involve get those mediawifi offline files, they even require a server farm :)17:09
kristianpaulin the other side there is humane project, who developed what seems an easy/fast alternative17:10
kristianpaulzrafa from #jlime ported it17:10
kristianpaulto nanonote's muffiman, i think is on qi wiki if you wanna try17:10
rjeffrieskristianpaul agree extraction the conent is a lot of work. but once done (maybe every 3 months or so) the value of access to wikipedia offline seems Most Excellent. OTOH the other offshoot of OM offers a purpose built off-like reader, and seems to not be thriving17:12
kristianpauls/qi/jlime17:12
rjeffriesI guess this is not all that interesting. fair enough.17:13
kristianpaulrjeffries: no work, is already there, well not using zim files but it is17:13
kristianpaulcheck jlime.com17:13
kyakwpwrak: thanks for sharing!17:28
rjeffrieskristianpaul as you prolly know, I do not own a Ben. But thanks. I don't need an offline reader for myself, but it seems like a cool idea maybe for kids?17:37
kristianpaulyes i'm aware  ;-)17:40
viricmth: what do you do with zram in the dingoo? Allocate 32MB of swap?19:26
qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.full_system-20120524-1615 22:05
mthviric: less than 32 MB, but we do use it as swap22:18
viricthe dingoo also has 32MB of ram?22:19
mthyes22:19
mthwe create a 20MB swap by default22:19
virichow much ram that takes?22:20
mthworst case 20MB of course, I don't know average case22:21
mtha lot of software doesn't swap at all22:21
mthAyla was the one who introduced zram, maybe he knows more22:22
viricok22:23
viricI remember I couldn't run 'g++' to build a simple program :)22:23
viricand I had to add some swap22:23
viricwell, it was swapping with the read only pages of the images22:23
viricbut I did not have any 'paging' for writeable pages22:24
qi-bot[commit] Werner Almesberger: change use of curly braces with "if-else" to kernel style (master) http://qi-hw.com/p/fped/424b74e23:32
qi-bot[commit] Werner Almesberger: inst.h (FOR_PKG_INSTS): protect against active_pkg == NULL (master) http://qi-hw.com/p/fped/b06ed9223:32
qi-bot[commit] Werner Almesberger: make active package track active variables (for the GUI) (master) http://qi-hw.com/p/fped/e0351bd23:32
--- Sat May 26 201200:00

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