#qi-hardware IRC log for Tuesday, 2012-01-24

ezdagorWhere do I get an "official" tarball?00:00
wejfrom my website: http://wejp.k.vu/projects/gmu00:00
Aylawej, you lost your 'p'?00:01
wejhehe, yeah ^^00:03
ezdagorWhere does it talk about the decoder? Are you referring to "decoder/mpg123" ?00:04
ezdagorwej: Are you referring to "decoder/mpg123" ?00:15
ezdagorOf course. That's what i'm referring to. Sorry for the typo.00:17
ezdagorThis is on the Ben, BTW.00:18
wejyeah, i figured00:18
qi-bot[commit] Werner Almesberger: m1/ledravaganza/placement.fig: illustrate LED placement (master) http://qi-hw.com/p/wernermisc/a44233e09:18
wpwrakhappiness is ... a new monitor, with working backlight :)17:52
viricwhen the lack of monitor is your biggest problem :)17:59
wpwrakmaybe not the biggest but a rather annoying one. well, i should buy some food, too. or else, i'll complain about something else before too long :)18:10
Artyomkristianpaul hello19:22
kristianpaulArtyom: hi !19:23
Artyomkristianpaul: I've thought a little about plans for the neasrest future touching osgps. If you are interested we can discuss them...19:24
kristianpaulgo ahead19:25
ArtyomNearest plans include: 1. Testing software with signal from sattelite (not from imitator).19:27
Artyom2. Replacing magic numbers by variables that must be calculated at program start19:27
Artyom3. Making simple signal generator in verilog. Actually this is just a simplified version of correlator channel that includes only carrier_nco, code_nco, code_gen and their control registers. (For example, original gp2015-correlator allows to use one of the channels for this purpose.)19:30
Artyom4. After this correlator channel can be tested in software without need of special signal-simulator or active_gps_antenna.19:31
Artyom5. Making two channel version (I'm afraid that more channels will not fit in FPGA that I use).19:32
ArtyomThat's it. What do you think?19:32
kristianpaulAbout 1. So this testing should be included in osgps? like a menu ask you to confirm signal acquisition from one satellite your know is in the skyview? 19:33
kristianpaulAbout 3. Looks nice too, i'll check this on gp2015 datasheet19:33
kristianpaulAbout 2. wich variables you mean? 19:34
kristianpaulABout 5. yeah two is a safe numer to grown more channels later :)19:35
kristianpauland later you may consider a bigger fpga (like the one in M1) to implement 12 correlators :)19:37
Artyom1. I keep in mind not exactly osgps port, but a simplified version which I use. (The idea to select a satellite at start is fine :))19:38
Artyom_2. I mean such values as carrier_nco control word, code_nco_control word, accum_int_periode control word and the reset which depend on the clock frequency that we use.19:40
Artyom_5. Yeah, later I will definitly use something more suitable for MM SoC ;)19:41
kristianpaulah, gotach you mean detect clock frequency, i like this so we can work with sige and maxim chips and may be others later19:42
lindi-just a quick note, are you all subscribed to foss-gps list?19:42
kristianpauli di19:43
kristianpaullindi-: but very much talk about rtklib no else... since i suscribed tought 19:43
kristianpauldam bot19:43
lindi-kristianpaul: exactly, you could post about your fpga ideas19:43
lindi-or Artyom_'s, I haven't yet read all scrollback19:44
kristianpaulyes please there is plenty of logs in this channel about it :)19:44
ArtyomI'm not sure that "to detect", but to set during compilation or something similar seems attractive19:45
Artyomno, I'm not subscribed. What is it?19:45
lindi-Artyom:   http://lists.osgeo.org/mailman/listinfo/foss-gps19:45
kristianpauldurin compilation yes, some header files should be enought :)19:46
kristianpaullindi-: you aware of opensource_gps mail list?19:47
kristianpaulvery quited to but osgps should be already been dicused on foss-gps too ,)19:48
lindi-kristianpaul: no, what's that?19:48
kristianpaulosgps project mail list19:48
lindi-"OpenSource GPS is software for x86 PCs that allows you to acquire, track and demodulate signals from GPS satellites."?19:48
kristianpaullindi-: http://home.earthlink.net/~cwkelley/19:49
lindi-( -- http://sourceforge.net/projects/osgps/ )19:49
kristianpaul"A starting point for learning about GPS with Open Source Software" 19:49
kristianpaulindeed :)=19:49
kristianpaulnot that rtklib or foxtrox are not important tought, but anyway19:49
kristianpaulmay be send some links invite people join here, check irc logs?19:50
lindi-kristianpaul: hmm, that site seems to have a lot of advertisements19:50
kristianpaulthere are jsut ideas and lots of works to do :)19:50
kristianpaullindi-: take your time, worth check it all19:50
kristianpaullindi-: if you we're aware of gps-sdr ospgs is same patch but no so blaoted :)19:51
kristianpauli mean C++ etc..19:51
lindi-kristianpaul: no, I've only used foxtrotgps and rtklib :)19:51
kristianpaulArtyom: i can try 1. and have soemthing for thursday19:52
Artyomkristianpaul: But you will have to change magic-nambers ;)19:52
kristianpaulah, yes19:53
kristianpauland i dont want to do the math by hand again..19:53
kristianpaulgood point19:53
kristianpaulArtyom: about osgps you consider using rtems for this, or try to keep use single apps no rtos?19:54
lindi-kristianpaul: so that is a webring? list of web sites?19:54
kristianpaullindi-: rtklib looks very interesting, just not have a fix yet here to complement position with it :)19:55
kristianpaullindi-: osgps you mean?19:55
lindi-kristianpaul: http://home.earthlink.net/~cwkelley/19:55
lindi-kristianpaul: I've done all my rtklib tests using openmoko gta0219:56
kristianpaullindi-: not at all check this http://home.earthlink.net/~cwkelley/introduction.htm19:56
Artyomkristianpaul: RTOS should be used anyway. RTEMS ssems attractive, but I didn't think a lot about it. May be some other RTOS, may be RTEMS. BTW Did you experiment with RTEMS?19:56
kristianpaulArtyom: RTEMS well yes, at least to make it work19:57
lindi-kristianpaul: ok, that seems to have last modified set to "2001", is this true?19:57
kristianpaullindi-: website yes, code have commitns some months old19:57
lindi-kristianpaul: does this osgps program have a git repository or something somewhere?19:58
kristianpaullindi-: yes! http://sourceforge.net/projects/osgps/19:58
kristianpaullindi-: the guy (cwkelley) Over the past 13 years working around GPS :-)19:59
kristianpaullemay too i guess check this  lindi-  http://en.qi-hardware.com/irclogs/search?q=lemay19:59
lindi-kristianpaul: and osgps is GPL? the files don't seem to have any copyright info20:00
kristianpaulahh good question it is, mom20:01
kristianpaul"The only requirement is that they are available free and clear under the GPL license."20:02
lindi-kristianpaul: also, do you need some proprietary assembler to compile the .ASM files? seems to be TASM syntax or something20:03
kristianpaulno no20:03
kristianpaulbut no for us20:03
kristianpaulthat was in the tiem osgps ran in a 133Mhz PC using a ISA card to talk gp2015 (Receiver)20:03
Artyomkristianpaul: If you have some experience with RTEMS then let it will be RTEMS ;)20:04
lindi-kristianpaul: it seems there are also binaries in the CVS20:04
lindi- http://osgps.cvs.sourceforge.net/viewvc/osgps/osgps/PCI.OBJ?revision=1.120:04
kristianpaullindi-: yes but ignore then :)20:04
kristianpaulwe dont use PCI20:04
Artyomkristianpaul: Is there any mailing list or something with osgps?20:04
kristianpaulArtyom: ye SS!20:04
Artyomwhat is SS?20:05
lindi-kristianpaul: I'm just looking at anything that'd block this from entering e.g. debian20:05
kristianpaulArtyom: typo sorry20:05
kristianpaulmy internet link getting slow20:05
lindi-some files have #include <dos.h> /* For Interrupt stuff */  still20:05
kristianpaulArtyom: http://lists.psas.pdx.edu/cgi-bin/mailman/listinfo/opensource_gps20:05
kristianpaulArtyom: Michele Bavaro is on that list too20:06
ArtyomMichele Bavaro ;) He mentioned my toy on his blog ;)20:06
kristianpaullindi-: from osgps for _now_ we jsut need gpsrcvr.c to make it work on our milkymist soc that uses namuru correlator20:06
Artyomkristianpaul: you too? ;)20:06
kristianpauldont think so, i havent even write it yet20:07
kristianpaulhe replied to me once i wrote osgps list20:07
kristianpaulabout using osgps implementation of gp2021 correlator20:07
lindi-kristianpaul: ok but I guess rtklib could be used as an alternative too?20:07
kristianpaullindi-: please tell us how :-)20:07
lindi-kristianpaul: if you can give me rinex from a single receiver I can calculate position easily20:08
kristianpauli dint found it very well documented, and it seemed to me i just work with certain navigation data formats?20:08
lindi-kristianpaul: I convert UBX to rinex using rtklib-convbin20:08
kristianpaulanyway we can *therically* provide that data, but no idea how rtklib could help from thre20:08
lindi-kristianpaul: it works here :)20:09
lindi-kristianpaul: so you don't have any raw data captured yet?20:10
kristianpaulif you mean raw I/Q data yes :)20:10
kristianpaullindi-: had you read about namuru correlator project?20:10
Artyomkristianpaul: you can process I/Q data with SoftOSGPS and have RINEX data as output ;)20:11
lindi-kristianpaul: not much20:11
kristianpaullindi-: can you point me an example where rtklib process a navigation message?20:11
kristianpaulArtyom: nope no :)20:11
lindi-kristianpaul: yes20:11
Artyomkristianpaul: why not? ;)20:12
kristianpaulArtyom: i event havent look at nmea code yet20:12
kristianpaulArtyom: i wodner if this work20:12
kristianpaulArtyom: you can process I/Q data with SoftOSGPS and have NMEA as output? :)20:12
kristianpaullindi-: please :)20:12
ArtyomI played a little this way. I recorded I/Q data with my front-end and processed it with SoftOSGPS. I received a position fix :) May be I should write a note in my blog about it...20:13
kristianpaulah yes, i tied that first20:13
lindi-kristianpaul: pntpos() in src/pntpos.c is afaik the simplest mode20:14
ArtyomOhnestly speaking: I didn't check nmea or rinex output. But I could see on the screen my coordinates20:14
kristianpaulArtyom: never got the fix.. but at that moment i dint knew it osgps dint support complex processing :/20:15
kristianpaulwell that told me cwkelley20:15
lindi-I do have access to a software defined radio but unfortunately I don't think I have time to capture you any live GPS data since that'd require some tinkering with antennas and amplifiers if I understand correctly20:15
kristianpaulno problem20:16
kristianpaulidea here is do live processing,20:16
lindi-rtklib can do live processing too20:16
lindi-but I've always done everything offline20:17
kristianpaulyes sure, but not in a risc cpu running at 80Mhz :)20:17
kristianpaulor yes?20:17
kristianpaulthats why the fpga and the namuru thing :)20:17
lindi-rtklib has been optimized to run on a beagleboard20:17
lindi-which is 500 MHz (?) ARM20:17
Artyomrtklib is interesting if you want to play with (don't know how it is exactly named in english) relative positioning. When you want to know coordinates of one receiver relative to another. In this case you can find relative coordinates very preciously. 20:18
kristianpauland prcecislly i heard?20:18
kristianpauli mean accurate20:18
lindi-yep, I got around 10 cm precision20:18
ArtyomWith OSGPS you can determine your coordinates with 5-10 meters accuracy. And with rtklib, like lindi mentioned, up to 10 cm20:19
lindi-http://lindi.iki.fi/lindi/finhack/finhack2010-rtklib-lindfors.pdf has some slides I made20:19
Artyomlindi: What are you working on? What is your project aims? ;)20:20
kristianpaulyes :)20:20
lindi-Artyom: that was just a crazy hobby project that escalated accidentally20:20
lindi-Artyom: I wanted to test rtklib and I had no raw data, at the same time I hated the fact that my openmoko ran non-free software to compute GPS position20:21
lindi-Artyom: so after some reverse engineering of the GPS firmware I was able to solve both problems20:22
kristianpaulArtyom: btw for osgps wich raw I/Q format you used 16-bit I/Q interleaved? and how much sampled time?20:23
Artyomlindi: So you used ANTARES chipset as your gps-receiver? (It's something like popular sirf?). And what data was processed by rtklib (rinex, nmea something else)?20:24
kristianpauli still want to record data, afaik i did wrong at the time i tried20:24
Action: kristianpaul feels bad about it20:24
Artyomkristianpaul: what kind of data do you want to record?20:25
lindi-Artyom: yep openmoko has ublox antaris 420:25
kristianpaulArtyom: the real out-put from my sige and your maxin board20:25
lindi-Artyom: slide 22 has links to the raw data20:25
kristianpaulArtyom: when i recorded it was 4 bit I/Q interleaved, then i have to bloat to 16 bits, the process with osgps20:25
lindi-Artyom: I asked the ublox chip to send me data in UBX format20:26
kristianpaulArtyom: but as i said this was before i realized osgps dint support real data20:26
kristianpaulArtyom: anwyay, jsut asking, that was 1 year ago.. i wasted too much time on it, then discover namuru and your blog ;)20:26
lindi-Artyom: slide 22 also shows how UBX is converted to RINEX20:28
Artyomkristianpaul: osgps as I remember works with real signals, not with I/Q-samples. So you have two possibilities: if IF-frequency of your front-end is high enough and sampling frequency is also high, then you can just through away Q-samples. Or you can hack OSGPS to work with I/Q-data. It's not difficult to make multiplication of complex-numbers (this is what I plan to add to namuru) ;)20:28
kristianpaulArtyom: with namuru is simpler i guess u simplify the correlator arm to I or Q, well i tought..20:29
kristianpaulArtyom: (multiplication) indeed but no now, :)20:30
kristianpaullindi-: yes osgps is not the most clean code, but i could bne ported to Milkymist SoC, ignore PCI, DOS and win stuff20:31
kristianpauleven softosgps :)20:31
Artyomlindi: ok, now I understand. And rtklib works mainly with RINEX?And you could archive decimeter-level accuracy with cheap reciever (working with L1 only band)? 20:32
Artyomkristianpaul: in correlator I use all 6 arms (Ie, Ip, Il, Qe, Qp, Ql).20:33
kristianpaullindi-: rtklib uses fft for live processing?20:34
kristianpaulah wait, it does not support real data samples right?20:34
lindi-kristianpaul: samples?20:35
lindi-kristianpaul: UBX has just results of the correlation20:35
kristianpaulyes i forgot .)20:36
kristianpaulUBX looks interesting format then :)20:36
lindi-see slide 1220:36
Action: kristianpaul downloads pdf20:36
Artyomkristianpaul: rtklib works with output of a GPS-receiver. For example it can potentially worj with OSGPS.20:37
DocScrutinizerlindi-: this pdf of yours still is awesome (though somehow triggers a deja vu sensation halfway thru ;-D ) - why don't you use adhoc wlan for data transmission?20:41
krispaulargh, blackout it seems20:41
lindi-DocScrutinizer: I just don't want to risk anything :)20:43
DocScrutinizernon-free software on 4th CPU? ;-P20:44
kristianpaullindi-: indeed, you can sell preconfigured openmoko with your soft, gps receiver with centimer accuracy are pretty expensive20:44
lindi-of wlan dropping some packets and compromising the data somehow20:44
DocScrutinizerduh, and the CPU in bq27200 also in proprietary ;-)20:44
lindi-doing these experiments wasn't exactly easy. I had to guard the receivers in a rather busy place to make sure nobody steals them :)20:44
lindi-anyways, this project is currently stuck until I can get better antennas20:45
DocScrutinizers/ in / is /20:46
kristianpaulArtyom: so.. you no rtems for now, you board dont  have the port20:46
kristianpaulArtyom: but may be i can run some basic code20:46
kristianpaulif memory map is same..20:47
kristianpaulanyway, will work on 1. :)20:47
kristianpaulWhat about you?20:47
Artyomkristianpaul: For now I will play a little with bare-hardware. I will make tasks that I have mentioned and then i will try ti switch to RTOS. A good news is that a port that I use should work with RTEMS. I even have hello_world application. But I didn't test it yet. May be next month...20:49
ArtyomAnd of course it will be greate if you will have some experience with RTEMS. Then we could progress faster as RTEMS will definitly slow down my work20:50
Artyomit;s time to sleep for me now...20:51
Artyomgood night20:51
kristianpaul15:51 still here :)20:51
kristianpauln8 !20:51
Artyomand 00:51 here ;)20:51
kristianpaullindi-: do you know about Milkymist soc?20:52
kristianpauland milkymist one?20:52
kristianpaulI would like make clear to you what are we trying to do20:52
kristianpaulbut you know what the nanonote is right?20:53
kristianpaulthe pocker computer with the a single chip20:53
kristianpaulxburst etcc.20:53
kristianpaulsorry viric !20:53
kristianpaullindi-: short words (taking words from your slides), we are implementing similar task that the ataris4 gps receiver from atmel does20:56
kristianpaulinside a fgpa20:56
kristianpaulmixing free hdl code from milkymist project, re-using work from namuru correlator (DSP part), and creating a firmware re-using osgps source code20:56
kristianpaulthe later output will be either a NMEA out (or rinex.. etc..) and of course a position fix20:57
kristianpauli dont know if you get this inresting or not, just wanted to tell, may be you want to tell same in the floss-gps list, i dont know if not having a devices that provide a fix is worth to mention in those lists :)20:58
lindi-kristianpaul: you still need non-free compiler to build all the fpga stuff?21:10
kristianpauland yes21:11
kristianpaulbut there are WIP alternatives like LLHDL21:11
DocScrutinizerkyak: skynet/guest4711 queried me "WHO THE FUCK ARE YOU?" :-D21:23
DocScrutinizerI honestly didn't know what to answer ;-)21:24
qi-bot[commit] Werner Almesberger: m1/ledravaganza/shine-through.fig: trim PCB layers for extra LED visibility (master) http://qi-hw.com/p/wernermisc/b13b4e822:59
wpwrakwolfspraul: did you pick -DKICAD_TESTING_VERSION=ON or -DKICAD_STABLE_VERSION=ON ?23:53
--- Wed Jan 25 201200:00

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