#qi-hardware IRC log for Thursday, 2011-07-21

GitHub134[milkymist] kristianpaul pushed 1 new commit to gps-sdr-testing: https://github.com/kristianpaul/milkymist/commit/6dd7b0ef4806c85e8683c79634fb06cc82944ea000:49
GitHub134[milkymist/gps-sdr-testing] Unified signal crossing sync by single module implementation and proper instatiation in top module - Cristian Paul PeƱaranda Rojas00:49
kristianpaulwpwrak_: what's the MCU or data bus specs i must find in a TFT display so i can say is VGA compatible?01:06
kristianpauli heard many from you even we can use the actual LCM bus from the nanonote to drive a LCD display but i never undertood why01:07
kristianpauldamn, namuru is full of combinational logic... and sebastien said is not good mix it with sequential logics, and the result of  untrackable bugs...01:16
kristianpaulokay lets start fixing the namuru reset :901:17
kristianpaulxiangfu: hello01:17
xiangfukristianpaul, hi01:17
kristianpaulxiangfu: there is Makefile or script that flash my nanonote from a "nightly" build?01:17
xiangfukristianpaul, I have one. I will try to add that to reflash_ben.sh as a parameter. just wait one memont.01:19
xiangfuif you want it now. I will send it to you by email01:19
kristianpaulsure, i think a lot make public, it will help a lot for your call for testing request before stable releases01:19
kristianpaulwell, i think01:19
kristianpauladd it to reflash_ben.sh as a parameter, will be nice, i'm not in a hurry to test nightly build but was thinkking how to make easier that process01:20
kristianpaulso take your time :)01:21
qi-bot[commit] Xiangfu Liu: nanonotefile: reflash_ben.sh support download and flash dailybuild images (master) http://qi-hw.com/p/openwrt-packages/b3c207401:30
xiangfukristianpaul, ^  :)01:30
xiangfudownload this file is ok: http://downloads.qi-hardware.com/software/images/NanoNote/Ben/reflash_ben.sh, 01:32
kristianpauli'll try01:32
xiangfufor example: ./reflash_ben.sh -d openwrt-xburst.full_system-06212011-102601:33
kristianpaulgood, that will make testing easier for lazy folks as me for sure ;)01:35
xiangfukristianpaul, maybe we also create a symlink after nightly build :)01:53
wpwrak_kristianpaul: (vga) hmm, anything that lets you generate a pattern of hsync, vsync, and "raw" video data, with the video data at least 4 bits wide02:18
kristianpaulqi-bot: raw video data is RGB?02:19
wpwrak_depends on what you do with it :)02:20
wpwrak_if you want to build something like ubb-vga, you'd just make a very simple analog mixer with a static RGB mapping (e.g., 3+3+2 RGB, or such)02:21
kristianpauli was asking you this because i feel i missing vga port from milkymist, due i dont have space for a second display all time, i was thinking in something smaller and came to my ming the  giantplus lcm i got for SIE wich is not doing nothing usefull right now02:22
kristianpaulbut i think drive it is not so out of the box VGA to LCM .. :_)02:23
kristianpaulof course i'm guessing :)02:24
wpwrak_lcms are pretty much little vgas. they have hsync, vsync, etc.02:25
kristianpaulbeep now make audible beeps on the nanonote02:59
wpwrak_kristianpaul: perfection has been achieved :)02:59
kristianpaulyeah :-D03:00
kristianpaulwow, even a different beep when tab autocompletion !03:02
kristianpauljust missing a RTC Alarm Clock and i can ripoff my cellphone ;-)03:03
wpwrak_perhaps the more important question is: how easy is it to turn it off ? :)03:03
wpwrak_yeah, it's a real shame that we still don't seem to have that03:04
kristianpaul(turn off) is not very loud, but i think it should be same as you did with mmc03:12
kristianpaularghh, i revert to a know state and got freeze again, i hope is not a cable issue..03:13
Action: kristianpaul turn on scope03:13
kristianpaulargghh not my day, screen bugging again...03:16
Action: kristianpaul after several a bit more agresive nocks work again03:18
wpwrak_kristianpaul: maybe this is old news, but do you know that Skyworks Solutions have bought SiGe ?03:43
kristianpaulwpwrak_: yes i'm aware03:45
wpwrak_good :) just stumbled across it03:45
kristianpaulthen i swiched to real mode :-)03:45
wpwrak_do you think the chips will be discontinued ?03:46
kristianpaulwell, i think thats a question for wolfgang, i really dont know, i'm aware of few gps devices using this chip, well, i should not be aware anyway..03:47
kristianpaulbut among hobyst maxim seems more popular afaik,03:48
kristianpaulso i really dont know03:48
kristianpauland if this happen, i'm not afraid, i mean i know what do, my problem right now is not wich fronted i have actually :-)03:48
kristianpauloops, wrong word :)03:49
kristianpaulthey have like 4 chips, for gps stuff, the popular seems 4110 acording to google, dunno if still produced03:50
wpwrak_but you switched because they were acquired ? or was that just an excuse to switch, because you didn't like their chip anymore ?03:51
kristianpaulno i like the chip03:51
kristianpaulbut wolfgang told me, well that due my slowness in this process will kinda hard to talk with then again03:52
kristianpaulso i tought a more generic aprouch will be good03:52
kristianpauldespite that, yes i like the chip03:52
kristianpaulhaving a I/Q downconverter, is really nice for software based processing03:53
kristianpaulalso in hardware, i mean correlator side, it make easier the process as it requires less sampling03:54
kristianpaulwich is not what happen in real mode..03:54
wpwrak_ah, i see03:54
kristianpaulnot only less sampling, also when you know in wich arm (I or Q) is the C/A code, well, you just stick on that correlator arm03:55
kristianpaulat least you want to break Y code as well ;-)03:56
kristianpaulso yes, SiGE chip make thinks lot easier, i'm not agains that, no :-)03:56
kristianpaulalso that rate in theory we can implement all process already in the nanonte, well i'm not doing that anytime soon, but just saying is posible if comples support to osgps is added03:58
kristianpaulwich is not that hard, for people already know how to do complex multiplication in C, wich is the missing part i remenber..03:58
kristianpaulso NO, i LIKE the chip (just for th record)03:59
kristianpaulokay, seems clk clock okay, so i did something to break again the wishbone CC bridge...04:01
wpwrak_(chip) now i'm curious why wolfgang made you switch04:01
kristianpaulno no04:02
kristianpaulhe not made me switch04:02
kristianpauli just warm/notice me about the issue of getting this chip for SiGE later..04:02
wpwrak_and that's why you switched04:03
kristianpaulwe're many reasons04:03
wpwrak_what's the issue he predicted ? that skyworks may kill the sige product line ? or that sige are in general hard to source ?04:04
kristianpaulone, i realized osgps dint support complex processing. so i wanted to do offline sampling04:04
kristianpauldont know04:04
kristianpaulalso i wanted to test SoftGNSS matlab code, wich now run in octave, but is not my  goal04:04
kristianpaulnamuru is ported, now is time make it work04:04
wpwrak_well, what was "the issue of getting this chip for SiGE later" ? ;-)04:04
kristianpaulthat the gave us a EVB, and we dont reply soon with results04:05
kristianpaulEVB was FREE also04:05
wpwrak_okay, so you're saying that Q/I is superior in theory but you lack the tools/knowledge to use it, and that's why you switched to real (and a different chip) ?04:06
kristianpaulas i was told all this companies want results in ie, 6 moths or something..04:06
kristianpaultools mostly, and yes knowledge to implement a software processing04:06
wpwrak_bah, i think they'll be happy whenever you show up as a customer. don't worry about being slow - much worse things can happen ;-)04:07
kristianpaulin namuru will be easier implement I/Q as it already do complex mixing/multiplication, then only missing and easy part  (i think) will be add support for a I and Q input to the correspoding I/Q arm04:08
kristianpaulis not hard, really04:08
kristianpaulbut in osgps well. i still not know all the source code04:08
wpwrak_(namuru) so they implemented complex math processing even if they don't use it ? interesting :)04:09
kristianpaulno no04:09
kristianpaullet me explain,04:10
kristianpaulthere are two mixing process when working with gps signals.04:11
kristianpaulthe first agains a carrier nco04:11
kristianpauland the second agains a local generate replica of C/A code04:12
wpwrak_the 2nd is the correlator, right ?04:12
kristianpaultracking, yes04:12
kristianpaulwell, forgive if i confuse terms, there are lot of them04:13
Action: wpwrak_ feels unusually bright ;-)04:13
kristianpaulso acording to Clifford Kelley, the complex mixing is not implemented in osgps04:13
wpwrak_gps is a bit of black magic to me, so don't worry ;-)04:13
wpwrak_the complex mixing would be in the 1st step ?04:14
wpwrak_or in both ?04:14
kristianpaulno, second step04:14
wpwrak_oh, interesting. i would have thought that this one was digital04:14
wpwrak_basically matching bit patterns04:16
wpwrak_the kind of thing you'd do with a state machine04:16
kristianpaulhum, well there is a step in wich you need to match some "magic bits" but thats after you confirm the C/A code signal, wich is basically mulutplying/mixing the local generated code with the data coming from first step i pointed above04:19
wpwrak_ah, so the "gold code" stuff is after the 2nd step ?04:21
kristianpaulwell, my understanding so far, i may miss something other technique for doing this for sure04:22
kristianpaulalso, i dont get get how the osgps do correlation as i tought was same as namuru but seems not at all04:23
kristianpaul"As far as I can tell the next major change you will need to make is to do a complex mixing of the signal with the sine and cosine functions." <- lifford Kelley04:24
wpwrak_sounds promising :)04:24
kristianpauli think osgps implement a software PLL, but i'm totally ignorant in that field, besides the analogy of the car used in racing to slow donw the whole race.. or something like that04:26
kristianpaulalso it is the FFT aprouch, as namuru is more the serial tracking i rememnber04:26
kristianpaulbut implement FFT i hardware is madness and unucessary when you have plemnty of counters04:27
wpwrak_from a cad system: "Cadmium has no GUI, which makes it very light weight and portable. " yes ! down with those pesky goo-eyes ! ;-))04:27
kristianpaulnice :-)04:28
wpwrak_(car used in racing) a pacer ?04:28
kristianpaulthe one tha slowdown other cars in case of an amergency04:29
wpwrak_i'm a bit surprised that there would be so much variety of what you can get from the RF frontend. each system seems to have a radically different approach to decoding gps, and none does what you actually need ...04:30
wpwrak_(cars) the one laying upside-down, burning, on the asphalt ? ;-)04:31
kristianpaulwell, i must PUBLICALLY said (i hope wolfgang read this), i had consult maxim chips literature is more rich than sige one, at least what you publically can get04:31
kristianpaullike ADC specs...04:31
wpwrak_(that cad) http://jayesh3.github.com/cadmium/04:32
kristianpaulpython vesion of openscad?04:32
kristianpaulah, no opencascade04:33
kristianpaulno car burning04:33
wpwrak_there are also python thingies for openscad. i think there are now two or three competing ones ;-)04:34
kristianpaulanyway, sogps uses a software PLL, that works with only real samples it seems.04:34
wpwrak_(sige documentation) yeah, sige suck quite a bit in that regard04:34
kristianpaulwas based for sige 4110 wich is a old sige chip04:34
kristianpaulDvdkhlng !! cheers for allegro games04:35
kristianpaulnice :-)04:35
wpwrak_(cad) there are even crazier things out there: https://christopherolah.wordpress.com/2011/07/16/surface-oriented-cad-math-telescopes/04:37
kristianpaulallegro alex4 game fix really good on nanonote screen, nice example04:37
kristianpaulwell, i stick on heekscad for now, and later... check that scriptying engine,04:38
wpwrak_(non-interactive cad) here's another one, with haskell bindings: http://hackage.haskell.org/package/mecha/04:41
kristianpaul(pacer) yes, this one https://secure.wikimedia.org/wikipedia/en/wiki/Safety_car04:42
wpwrak_and now the happy python-to-openscad bindings: a recent one: https://github.com/hmeyer/openscadpy04:42
kristianpaulthats a PLL for me now 04:42
wpwrak_pyscad: https://github.com/kevinmehall/pyscad04:46
wpwrak_and pyopenscad: https://github.com/etjones/MCAD/tree/master/PyOpenScad04:46
kristianpaultoo many py*cad, is crazy!04:47
wpwrak_just wait for the perl, ruby, .. bindings :)04:49
kristianpaulzear: man allegro is awesome for games, check this http://www.allegro.cc/depot/demo/04:49
kristianpaulzear: and already run on the nanonote ;), just two demo games ported so far, i know you can beat that ;)04:50
wpwrak_(cad) and then there's lolcad: http://diyhpl.us/cgit/lolcad04:53
kristianpauloh, Bryan Bishop04:54
kristianpaulhe was working on some py*openscad bindings i remenber04:55
wpwrak_heh, don't ask me about people. i have a lousy social memory ;-)04:56
wpwrak_but the only pyopenscad committer seems to be etjones04:57
kristianpaul(people) nah. just the name still on my memory because he was working on apt-get for hardware 05:03
wpwrak_apt-get for hardware ? apt-get teapot  ? :)05:04
wpwrak_s/ /install/05:04
kristianpaulsomething out-of the box from you computer to the cnc/3d printer i think05:09
kristianpaulno idea status of this,05:09
wpwrak_yeah, the path from design to STL now seems scriptable. but from there on, there's still too much manual stuff (using HeeksCAD)05:11
wpwrak_and it's actually that latter step where scripts would be most useful05:12
wpwrak_e.g., for adjusting the workpiece position, the area in which to work, corrections for mill tolerances, etc.05:12
wpwrak_python-occ is quit a fat beast. 41.5 MB05:13
wpwrak_but works :)05:17
kristianpaulthan plan, of taking naps in the evening, is really interesting :-)05:59
wpwrak_there are a lot of things we can learn from cats - take a nap whenever you can, stay up all night, hunting :)06:00
kristianpaulhum, i never got back ack signal from the core, that explain the freeze and the unhandled expection !06:28
kristianpaulbah, 3hrs nap :p06:54
kristianpaulbtw nice website http://www.examplenow.com/tee/06:57
Action: kristianpaul just discovered tr command06:58
larschm... something messes up my stackframe... this is annoying06:58
kristianpaula rootkit perhaps? 06:59
larscnah, broken code ;)07:00
larscsyscalls in general work fine, but when using execve the stackframe is messed up afterwards07:01
larscthough execve from kernel space works fine07:02
kristianpaulha, yes i have cyc signal but never ack (from the master point of view)07:02
kristianpaulwell, something is not right with the wishbone CC bridge... i'll  see07:03
kristianpaulbut i was..07:03
kristianpaulfor the future, not let sinthesizing a bittream and after a day see what was suposed to do.. :-S07:04
kristianpaulindeed, jon just jump with questions some how07:08
kristianpaulbtw larsc you mentioned networking is working with uclinux for mm1, that include last minimac2?07:08
kristianpaul(already asked by ML too)07:08
larsckristianpaul: only minimac107:11
kristianpaulthats history...10:45
larscnot on my milkymist ;)10:53
larschow is minimac2 different?10:55
kristianpaulyeah, no DMA basically, replaced by a 2k  FIFO i remenber10:55
kristianpaulall this because some memory bugs presented in the previos core10:56
methril_worksomeone from uk?11:48
wolfspraulmethril_work: zedstar_ is11:49
methril_workthank you wolfspraul :)11:49
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-07202011-0935/13:03
wpwrakinteresting .. i've been running without swap for the last months23:34
--- Fri Jul 22 201100:00

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