#qi-hardware IRC log for Monday, 2011-08-01

kristianpaulah is true http://gnss-sdr.ru/media/1/20101004-screenshot.gif, Artyom dint get track..00:25
kristianpaulyeah, same situation as my self :-S00:25
kristianpauli guess osgps never got track as i was feeding it with complex data00:25
kristianpauleven after sample 40Mb of data00:25
kristianpaulalso this gps-sdr from CTA es kinda slow and cpu consuming, was very wise from him to swtich to scilab :-)00:27
wpwrakso what's the plan then ? wait for Artyom to fix things on his side ?00:31
kristianpaulno of course not00:32
kristianpaulalso his side have its complexity that he only understand00:33
kristianpauli still on my namuru debugging, afaik cant make carrier nco to work..00:33
kristianpaulonce i get it work, the accumulator should start to work00:34
wpwrakah, so your troubles are still in earlier processing stages than his. i see.00:35
kristianpaulwith namuru yes00:35
kristianpaulbut in full software processing kinda same 00:35
kristianpaulbut i had same issues as heen before with the acquisition module, i got prn code detection but couldt lock it..00:37
kristianpaulthat was with osgps00:37
kristianpauli soft mode to put it a name00:38
kristianpaulhe was very kind to sent my hist vhdl namuru code00:38
kristianpauli discover i missed a state for the time base counter on my verilog code00:39
kristianpaulfixed now, i'll try as soon sitesis finish00:39
wpwrakcool. that's probably a good bug to fix ;-)00:40
kristianpaulwell, i depend may be xst asumed what he wrote in the code and i specify00:42
kristianpauland sadly it wast related to my carrier nco stuckness...00:42
kristianpauland i'm using carrier_nco code as it is..00:42
kristianpaulso the error should be in higher layers..00:43
wpwrakthe harder the bug to fix, the greater the glory ;-)00:48
qi-bot[commit] Werner Almesberger: ptrude/: functions for path stretching and calculation of proportional length (master) http://qi-hw.com/p/cae-tools/7ef626900:55
qi-bot[commit] Werner Almesberger: ptrude/: changed extrusion from xy path to xz path; variable number of faces (master) http://qi-hw.com/p/cae-tools/ee6344500:55
qi-bot[commit] Werner Almesberger: ptrude/: many major math fixes, especially in stretch_path (master) http://qi-hw.com/p/cae-tools/26a0f4c00:55
wolfspraulkristianpaul: don't you think the Nyan cat is more like 400 microns high?07:28
kristianpaulwolfspraul: dunno may be azonenberg can tell us more11:49
Artyomhi kristianpaul11:56
kristianpaulhello 11:56
ArtyomI've seen discussion about real/complex data samples from your front-end11:57
kristianpaulah yes,11:57
kristianpaulit make complex out of the box..11:57
kristianpaulbut at time i tested i havent software avaliable for complex procesing.. :(11:58
Artyomand what is sampling frequency and intermediate frequency?11:58
Artyomand bandwidth?11:58
kristianpaul2.556 intermdiate 2.048Mhz sampling11:58
kristianpaulevery sample 4 bits11:58
ArtyomBut can you use higher sampling frequency?11:59
kristianpaulyeah, in real mode11:59
kristianpaul16.694Mhz of sampling11:59
kristianpaulsame IF, 2 bits per sample12:00
kristianpauli switched to that mode now12:00
kristianpaulbut all went i can try complex agains namuru12:00
kristianpaulbtw i checked you code, i added a codition to the counters that i missed, but of course time_base was working before12:01
kristianpauland i still get the carrier nco stuck :(12:01
ArtyomIn my front-end I also have complex-data output. I have IF=2.42MHz Sampling=16MHz and bandwidth=4.2MHz. When I work with osgps, I just throw away all Q-samples and use only I-samples.12:02
kristianpaulArtyom: every sample 4 bits at 2.048 MSPS at 8.192 Mhz sampling clock in other words12:02
kristianpaulstill using maxin right?12:02
kristianpaulyou trought away all Q-samples?? but what happened if carrier mas in that?12:03
kristianpauli may be wrong but i think is no always exactly this guess12:03
Artyomwhat do you mean "what happened if carrier mas in that?". Can you explain in more detail...12:05
kristianpaulah, moment you work with GLONASS not GPS right?12:05
kristianpaulin GPS L1 signals there is a P-code and C/A code 12:08
ArtyomI work with both. But now as I use osgps+namuru I work with GPS only. Later It will be easy to switch to GLONASS12:08
kristianpaulon is in phase other quadrature, (dont remenber wich on is who..)12:08
ArtyomYes, In GLONASS the same thing12:08
kristianpauloh nice12:08
kristianpauli may may be wrong but i undertood that processing just Q for example you cant get the P-code12:09
kristianpaulbut as i said, is a _guess_ for me i never dig in to12:10
ArtyomNo, this is wrong12:10
ArtyomIn receiver there is another meaning in quadrature channels12:11
ArtyomWe use quadratures in receiver when we work with signals with unknown phase (like it is in GPS/GLONASS/COMPASS/GALILEO)12:12
kristianpaulwell. i'll check more theory about this later12:15
kristianpaulbtw before i leave cause i travel to work like 2hr and i have a meeting in the morning..12:16
kristianpaulthe carrier phase register in namuru just need to be written once isnt? well at least for initilization, later i may be ajusted acording to doppler etc..12:17
kristianpauli think i'll write a testbench for it, and debug it individually..12:18
Artyomwhich exactly register are you asking about?12:18
kristianpauland  CARRIER_NCO12:18
Artyomduring acquisition you set CODE_NCO to a fixed value12:19
Artyomand CARRIER_NCO is changing according to current doppler-bin (where we search the signal)12:20
ArtyomAnd during tracking we always change their values12:21
Artyomboth CODE_NCO and CARRIER_NCO12:21
kristianpaulbut then CARRIER_MEASUREMENT will not change its value over the time if i dont change the CARRIER_NCO ?12:23
kristianpaulcause i'm initializing both CODE_NCO and CARRIER_NCO with fixed values12:24
kristianpaulif i read CODE_MEASUREMENT it looks like is changing vauet but not the case for CARRIER_MEASUREMENT..12:24
kristianpaulthat let me thing something wrong was happening with the carrier nco module..12:25
Artyommay be, I can't tell you exactly as I didn't dig so far. I only tried to control tracking loops. And CODE_MEASUREMENT and CARRIER_MEASUREMENT are used later for calculating pseudoranges... (if I'm not wrong)12:26
kristianpaulhum. but how you track if the carrier nco is not fixed first?12:31
kristianpaulor i'm misiing something..12:31
kristianpaulagrh, i must leave now, but my irc client gets here catching12:32
kristianpaulbye Artyom 12:32
kristianpaulbut you read the accumulators right? i mean after initializr CODE_NCO and CARRIER_NCO ?12:34
kristianpaulalso if read it must be before got overwrite 12:35
ArtyomI read only i_prompt,q_prompt,i_late,q_late12:35
kristianpauli read all, but zero is the cotent tought..12:35
kristianpaulokay now leaving, i'm late ;)12:36
ArtyomI must make a note: I tried to use my own code for tracking. In general we have only to read i_promt,q_prompt,i_late,q_late_i_early,_qearly in order to calculate errors of our carrier-generator and code_generator. And based on these 6 values we can calculate new values for CODE_NCO and CARRIER_NCO.12:38
ArtyomIn osgps there are some additions to the general rule:12:39
ArtyomDuring pull-in process additionally navigation bits border is also identified. And additional checks are made in order to be sure that the process is working correctly.12:42
ArtyomAnd also osgps uses only 4 values i_prompt, q_prompt, i_late, q_late.12:43
ArtyomBased on these 4 values every ms during pull-in process new CARRIER_DCO and CODE_DCO values are calculated. During tracking process CARRIER_DCO is updated every ms and CODE_DCO is updated every 20 ms.12:49
ArtyomAnd CODE_MEASUREMENTS/CARRIER_MEASUREMENTS are used only every 100 ms to calculate pseudoranges.12:50
qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-07312011-1105/14:24
rjeffrieswolfspraul I was w-r-o-n-g re ARM costs. The use case I was thinking of is this part:14:25
rjeffriesNXP LPC1768, ARM Cortex M3 100MHz, ~$5 in >1K volume.14:25
rohmeh.. this shipping logistics makes my brain hurt14:26
rohthese madmen who wrote those texts and regulations14:27
rjeffriessemi-unrelated factoid: the per-piece royalty to ARM when a licesee produces an ARM-based CPU is as low as $00.07 per chip 14:27
rjeffrieshello roh14:27
rohrjeffries: yeah. thats not really a cost factor for us at the volumes we do.14:28
rjeffriesroh I was not suggesting it was. I had calimed a 1GHz Arm was $5 I was wrong about that14:28
wolfspraulrjeffries: fair enough.14:29
wolfsprauleven lower than 7 cents14:29
wolfspraul2-40 cents, afaik14:29
rjeffriesit is a tough business for sure14:30
rohimx233 is 7.05Euro at digikey14:30
wolfspraulnah, all fine. they don't shoulder the expensive and risky part.14:30
rohwolfspraul: any idea what zolltarifnummer to use?14:36
wolfspraulno idea14:36
rohfsck.. i havent got a clue either.. there are thousands14:38
rohit seems i have to atleast fill a CN2314:47
rohfsck. need this number.. gah. 1h till the shop closes14:53
wolfspraulroh: is that the same as HS code? I found "390690  Other Acrylic Polymers"15:09
wolfspraulbut there's also metal and other stuff inside...15:09
rohexactly.. i only found codes withing 39xxxxxx which are for uncut material15:11
rohthis zoll-lists are a serious case of TL;DR15:19
wpwrakroh: try to google for something similar in china. they sometimes list the HS code.15:45
wpwrakroh: i.e., acrylic case screws "hs code"15:45
wpwrakfirst hit: Face plastic case with screws (HS Code : 85439000, made in china)15:45
wpwrak854390  "Parts of Particle Accelerators, Audio Mixers, High Frequency Amplifiers" hehe ;-)15:47
rohmaybe i should just note the first 4 chars and a ?15:48
rohneed to run. closing in 10 minutes15:48
wpwrakyou can abbreviate the numbers. but i don't remember to what point.15:49
wpwrakask them if they actually want the code :)15:49
wpwrakfedex usually do (but they're probably okay without it, too)15:49
wpwrakwhen i asked at the postal office, they said "no" (there's room on the form for it, in case you insist)15:50
wpwrakwolfspraul: how's the news coming along ? the 2 months schedule seems to be a little tough for the sheer quantity of things - even if july was actually fairly quiet on the ben side (and quite the contrary on the m1 side)16:20
rohshipment shipped16:56
wpwrakdid you put a code ?17:08
rohfilled a CN2317:10
LunaVoraxhello everyone !18:38
LunaVoraxLong time no see !18:38
LunaVoraxWhat's up since the last month ?18:40
--- Tue Aug 2 201100:00

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