#qi-hardware IRC log for Saturday, 2010-11-27

ss547i had problm with my pc00:33
wpwrakheh, this time i found the piranha ratio - 3 rounds of etching in just 20 minutes. watch this, iron(III) chloride ! :)02:50
wolfspraul:-)02:56
viricwpwrak: I'd like to learn to make my own PCBs by chemicals... can you recommend any book about that, covering different techniques?04:27
wpwrakviric: not sure if there are books ...06:57
wpwrakviric: this document has some of the process i used a few years ago: http://www.almesberger.net/misc/led-toy/ledtoy-building-0pre9.pdf07:00
viricthank you!07:03
viricI'll read it.07:03
wpwrakviric: what has changed since: i usually cut pcbs with a CNC mill, i use a different acid, i usually use a laminator for toner transfer instead of an iron, i now use lead-free and better flux (different temperatures, different cleaning process, etc.), i use smaller components (different process)07:04
wpwrak(smaller comp) well, different soldering technique. depends a bit on on the type of component. back then, i thought SSOP and 0603  were the smallest i could handle :)07:05
viric:)07:07
wpwrakviric: ah yes, and the dual-sided process has a variant now that's more accurate but also slower :)07:08
viricand you learnt from where?07:08
wpwrakthe internet and experimenting :)07:08
viricaha, ok07:08
viricI have some friends who master some of those techniques07:08
viricI'll also ask them07:08
wpwrakthe toner transfer method is explained in great detail on the net. the reference is in the document07:09
wpwrakah, another process variant: instead of acetone, which can be difficult to get, i now use paint thinner most of the time07:10
viricah ok07:13
viricacetone is difficult to get nowadays?07:13
viricI did not know07:13
viricDoes anybody know of a voip phone with open hardware?07:26
viricnot a mobile phone07:27
kristianpaulviric: close i know in mesh potatoe, not sure how open hardware is, but seems is a bit08:13
virickristianpaul: 'in mesh potatoe'?09:11
viricla malla patata?09:12
viricah I found about it.09:12
viricthe mesh potato looks great. thank you!09:21
viricwpwrak: I like a lot your ledtoy-building :) thank you very much!09:46
wpwrakyou're welcome :)09:48
kristianpaulIs this a fifo http://paste.debian.net/100897/ ?11:46
kristianpaulhow horrible is C++ sintax :(11:46
kristianpaulwell not so, but this ok keep in mind object oriented thing is something i'm not used to11:47
kristianpaulok.. a struct made of other struct.. http://paste.debian.net/100900/11:55
wpwrakeh ?11:58
kristianpaulhe i may be wrong, just gettting used to C++ codying style ,-)11:59
larscthats C ;)11:59
kristianpaulah?11:59
wpwraklooks pretty c-ish so far. you haven't touched the hells of C++ yet :)11:59
kristianpaulohh12:00
kristianpaulis not .cpp C++ related?12:00
kristianpaulwhat is the purpose of a struct in C, i tought i wast just used in OO programing12:01
viricthe purpose of a struct in C? :)12:01
wpwraka struct provides structure :)12:02
wpwrak"oo" is a sound programmers make when experiencing pain12:02
virichehe12:02
virickristianpaul: programming science developed some abstractions over linear memory, since some time ago12:03
kristianpaulyeah :), but.. is like like create a new type of variable?12:03
Action: kristianpaul reads wikipedia12:04
viricthere are even more abstractions beyond 'int'12:05
wpwrakbah, void * and __asm__ is all you need :)12:07
viricif talking about 'needs', yes :)12:07
kristianpauloh thre is a fftw implementation in asm? ;)12:07
kristianpauls/thre/there12:08
viricpart of the humanity (although little) have the chance to go beyond 'needs' tohugh12:08
kristianpauldamn i'm ignorant in C :/12:15
viricThat is a good start, for learning! :)12:15
viricDoes the nanonote 'reboot' well?12:17
viricI mean... I got "halp -pf" finally powering off12:18
kristianpaulhalp?12:18
virichalt12:18
viric:)12:18
virichalt -p does not power off though12:18
kristianpauloh i was scare for a while12:18
kristianpaulwhat about init?12:18
viricsysvinit12:19
viric# runlevel12:19
viric3 612:19
viricstarted at 3, now at 6.12:20
viric(notice dropbear still has my shell open)12:20
kristianpauloh is a collection12:23
kristianpaulbetter ilustrated in a cartesian plot12:23
viric:D12:33
viricit only shows the 'start' and the current12:34
kristianpaulaahh int16t is prefered because MMS/SSE use it15:34
kristianpaulben simd uses same?..15:35
kristianpaulhmm what happen if for now i just ignore Imaginary part..15:47
wpwrak_your bandwidth drops by 50% ?16:14
wpwrak_kristianpaul: but why do you want to optimize now ? have you already decoded some samples ?16:52
kristianpaulwpwrak_: no, i'm impplementing verilog module to be loaded ina memory buffer then dumped to the nand,17:25
kristianpaulok not optimize now :)17:25
kristianpaulwpwrak_: actually bandwich is same, i still usign an int8_t or int16_t17:27
kristianpauljust less bits filled17:27
wpwrak_(bw) what I mean is that, if you drop the imaginary part, this is equivalent to sampling at half the rate17:33
wpwrak_well, more or less equivalent, i guess :)17:33
wpwrak_do you already have suitable algorithms for processing Q/I data ? or do you have to design them yourself ?17:34
kristianpaulgps-sdr is my first guide17:42
kristianpauli just documenting my self to make data wasy to be intepreted ny this software17:42
kristianpaulwhoa ! i have a mail from then, since monday, (about some questions i made) lets see17:43
kristianpaulother similar projects (with closed hardware stuff) use it17:44
kristianpaulso i decide follow since at least the prove it that can be forked17:44
kristianpauljsut for this step in wich i just plan use SIE as front end to the SiGE EVB17:45
wpwrak_so the algorithms they have look usable to you ? i.e., they have the right data format and so on ? (or if they don't, you know how to convert between the format you have and theirs ?)17:45
kristianpaul(if they don't,) i find out how to format the data17:48
kristianpauli'm findig*17:48
kristianpaulwhoa looks this project, thats stuuf that google dint tell in the first page ;) http://www.gmat.unsw.edu.au/namuru/17:48
wpwrak_big board with big antennas :)17:50
wpwrak_ah o, they're shields. better ;)17:51
kristianpaulwell it was a link froma reply from one sdr-gps developer wich is aware i'm using fpga stuff17:55
kristianpaulah tricky why use 2 bits sign/magn if from the posible  4 results 2 have no sense (+0, -0) so the other SiGE chip just used 1 bit for phase and qautrautre !18:02
kristianpauls/qautrautre/quadrature18:02
wpwrak_0 does seem to convey information. +/- seems wasteful, though18:03
kristianpaulyeah but i ditn get that the week before, and was strugling because it18:04
wpwrak_so you basically have three deltas encoded in two bits: -quantum, no change, +qantum, right ?18:05
kristianpaulyes18:05
wpwrak_kewl, i understood something :)18:06
kristianpaul?18:06
kristianpaul:)18:06
kristianpaulno change is tricky (i dint get how will help me)18:07
kristianpaul:p18:07
wpwrak_do you feed the data directly into the FFT or do you integrate first ?18:12
kristianpaulgood question :)18:14
kristianpaulthat a BIG doubt i have18:14
kristianpaulFFT can handle of course real and imaginary data18:14
kristianpaulat least a book i have said that, but i'm not aware of how gps-sdr handle it18:15
kristianpauli'll ask that to the developer :)18:15
wpwrak_maybe try to find some example of anything that uses a sign/magnitude stream and see what they do with it ?18:16
wpwrak_could be that there's also some other kind of filtering or stuff you can do first. just integrating seems a bit wasteful. e.g., there may be useful phase information in the high-rate bitstream.18:17
wpwrak_there's also something called decimation. that's also a rate reduction. the USRP uses it quite heavily. (as in "you must decimate or it won't work") i haven't investigated that yet, though :)18:18
kristianpaulha i need calculate or get from somwhere a itnermediate frequency18:24
kristianpaulbut i just have filter centre freq and bandwich..18:25
kristianpaulok, there is a radio eng aroudn?18:25
kristianpaul:)18:25
wpwrak_would be kinda useful for me, too :)18:25
wpwrak_eventually, i figure out everything i need by myself, but a quick answer sometimes beats weeks of trial and error :)18:26
kristianpaulhope wolfgang read this log :)18:26
kristianpaulwpwrak_: (weeks of trial and error) totally agree18:27
kristianpaulwell i my case read read and confuse then re-read to getter the idea and read again,18:33
kristianpaulbut i'm close to start trail and error18:33
wpwrak_the read-confuse-read-... cycle is also a good one ;-)18:40
kristianpauls/trail/trial18:50
kristianpaulhttp://www.dynamics.co.nz/index.php?main_page=page&id=9  This fully functional GPS receiver design can provide 12 or more channels of parallel tracking hardware and software.19:02
kristianpaulwpwrak_: ^ well if the antenna is big looks good anyway :)19:03
kristianpaul12 channels is heacy my current closed source gps cant handle more than 8, with the test i made19:03
kristianpaulha Zarlink is SiGE competitor19:05
kristianpauli'll enjoy GPS reading this week :)19:06
kristianpaulhttp://www.zarlink.com/zarlink/hs/82_GP2015.htm <-  Application Notes :O19:06
wpwrak_monster circuits :)19:08
kristianpaulmosnter in wich sense?19:09
wpwrak_(zarlink) also sign/mag. but simpler protocol - no SYNC19:12
wpwrak_(monster) lot of stuff on these boards19:12
kristianpaulyeah i noticed no SYNC :/19:12
kristianpaullots stuff indeed :)19:12
kristianpaulha they sell a "GPS Receiver Baseband Processor"19:13
kristianpauli wonder why still usign to separate chips19:13
kristianpaulah got my answer here http://gps.psas.pdx.edu/19:18
kristianpaulmay be that the GP4020 chipset isn't going to be available for a long time19:19
kristianpaulquuoting^19:19
wpwrak_yeah. there may simply be no sufficient economical incentive to integrate everything into once chip. depends on their customers and technical capabilities.19:20
kristianpaulhttp://oscar.dcarr.org/ssrp/index.php19:31
kristianpaulNamuru V2 GPS receiver without IMU $4,620.00 USD !!19:40
kristianpaulWHAT?19:40
kristianpaulthen you get the sources19:40
kristianpaulha19:40
kristianpaulwell if is GPL/BSD anyw ay is okay they are not pushed to make stuuf public19:45
kristianpaulstuff*19:45
kristianpaul"--prefix=/usr" <. que peligro20:05
kristianpaulpara eso iventaron /opt ..20:05
kristianpaulen fin20:05
kristianpauloops20:07
kristianpaul(decimation) not for me i dont want deal with more downsampling processes..20:31
kristianpaul[the USRP uses it quite heavily. (as in "you must decimate or it won't work")] wow, i wonder why and in wich step is used20:32
kristianpaulmay be i wont work on some not so FAST computers?20:33
kristianpaulmissing 8 cores and 8Gb in ram? :)20:33
kristianpaulah wait20:35
kristianpaul"D set = decimate (1 bit instead of 2)"20:35
wpwrak_naw, the ADC produces 100 MSa/s, but you can only transfer about 25 MSa/s to the PC. that's why you need to decimate by at least 4.20:35
kristianpaulis what i commented before about 2bits are not needed20:35
wpwrak_1 Sample = 2 * 16 bits, so 25 MSa/s = 800 Mbps20:36
kristianpaulyes make sense20:36
wpwrak_decimation works on the complex value and still gives complex as a result. there's no escape ;-)20:36
kristianpaultell me more if you found something more about it, looks intersting20:39
kristianpauloh you meant hacked in the good sense?20:50
kristianpauloops20:50
kristianpauldamn irssi windows20:50
qi-bot[commit] Werner Almesberger: vialtray: make a tray for vials containing 0402/0603 components (in progress) http://qi-hw.com/p/wernermisc/1d72af720:55
qi-bot[commit] Werner Almesberger: libbb: library for bit-banging the Ben's uSD port (in progress) http://qi-hw.com/p/wernermisc/458702c20:55
qi-bot[commit] Werner Almesberger: vialtray: ready for first test run (on-going) http://qi-hw.com/p/wernermisc/649e00b20:55
qi-bot[commit] Werner Almesberger: vialtray: more parameter adjustments http://qi-hw.com/p/wernermisc/1d24b2e23:30
--- Sun Nov 28 201000:00

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