#qi-hardware IRC log for Sunday, 2011-07-31

kristianpaulArtyom G around?00:14
kristianpaulwpwrak: writing manuals?00:16
kristianpauloh man i want a funcube dongle http://www.oz9aec.net/index.php/gnu-radio/gqrx-sdr :-)00:17
kristianpaulplus a wearable antenna, all done ;-)00:18
kristianpaulha, long mail :)00:21
kristianpaulyeah, SigE is very good writing shor datashetets ;)00:22
wolfspraulrjeffries_: which manufacturer sells a 1ghz arm for < 5 usd ?02:46
wolfspraulmy point was a price comparison to fpga and I didn't want to mislead by using too low a number for the high-end arm asic02:46
wolfspraula quick check on digikey suggests that 15-20 usd is even too low, you'd have a very hard time sourcing even an 800 mhz chip at that price range02:47
wolfspraulin low volume, it may easily be 30 USD and more02:47
wolfspraullooking at Freescale ARM right now, for example02:47
wolfspraulnone of the 800 mhz or 1 ghz variants are in stock, phew02:49
wolfspraulhere's one, 800 mhz. http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=MCIMX503CVM8B-ND02:51
wolfspraulUSD 23.22 if you buy one, going down to USD 18.28 if you buy 10002:52
wolfspraulmy number of 15-20 USD for a 1 ghz arm was probably a bit low, for our quantities02:52
wolfspraulas with fpgas, there are many variants, prices may vary considerably depending on which features exactly you have in the chip. mhz is just one parameter of many.02:54
wolfspraullet me know what you have in mind for 5 USD :-)02:55
wolfsprauloh wow, I rarely look at ARM prices, but I am so surprised how expensive the TI OMAP are03:00
wolfspraul40-70 USD at digikey, even for the 500-600 mhz range03:01
wolfspraulno wonder mediatek and now spreadtrum are taking the market from the bottom03:02
wolfspraulI get two full Android smartphones for the price of one TI OMAP chip :-) (exaggerating, but it gets close)03:02
kristianpaultwo for the price of one, nice ;-)03:48
wpwrakkristianpaul: (short data sheet) well, to be honest, some of sige's material is more like an insult than a data sheet :) consider this page: http://www.skyworksinc.com/Products_SiGe_Smart_Energy.aspx04:26
wpwrakkristianpaul: there you have nice and proper for all the chips, except for the SE2436L, the only one with a built-in balun. now really, how does a bloody balun change it from something that's okay to tell everyone to a piece of information you have to beg them for ?04:28
wolfspraulwpwrak: I think there's a trend to go closed. What do you think?04:32
wolfspraulit's a bit hard for me to follow/judge with so many companies and datasheets. and we so radically focuse on those that do have full datasheets, even if it means to make an unusual selection or design.04:33
wolfspraulin the SiGe case, I had a meeting in Hong Kong already agreed to, but didn't want to go without some more real progress from kristianpaul. That was about a year ago :-)04:33
wolfsprauleventually we'll have that progress, then I see whether my email contacts still work there and still want to meet :-)04:33
wolfspraulafter that I know more about SiGe's IP and datasheet policy/strategy/etc04:34
wpwraki don't really see a trend. it's more like company-specific and sector-specific styles. e.g., wlan is usually closed, ieee 802.15.4 usually open, etc.04:34
wolfspraulyes, I would agree04:34
wpwrakthen companies like atmel, skyworks, rfmd are usually open while microsemi are closed and sige half-half (to pick examples from the RF front-end world)04:35
wolfspraulnew semiconductor startups are almost always very closed I found04:36
wpwrakyes, they have too much to fear :)04:36
wolfspraulonce you get to the level of passive components or other 'low level' semiconductors, it's mostly open04:36
wpwraknot enough money to hire mercenaries ;-)04:36
wolfspraulonce you move into anything baseband/rf, it gets very closed again04:37
wolfspraulit's like the baseband lives in a parallel reality or something04:37
wpwrakyup, complexity is a factor04:37
wolfspraulno it's not just C code running in a core, it's 'magic'04:37
wpwrak;-)04:37
wolfspraulso much fuss about basebands and dsp, really04:38
wolfspraulmostly closed, I'm constantly looking for more documentation on such chips, but hard to find04:38
wpwrakwell, there's an active patent war in that area, so ...04:38
wpwrakparticularly as you move up to the higher performance items04:38
wolfspraulCMOS image sensors, also very bad04:39
wolfspraulmaybe the same for those radar chips Sebastien is talking about sometimes?04:40
wolfspraulwe should have a table, in which industries/areas one is likely to find high quality open datasheets, or which companies are known to have open/closed policies04:41
wolfspraulthe problem is that the judgment is so subjective04:41
wpwrakheh, that could be a nice "map"04:42
wpwrak(judgement) you the company's judgment whether to be open or close, or ours whether they are ?04:42
wolfspraulours whether they are04:44
wolfspraulactually your mail, as so many times, set a good standard04:45
wolfspraulmake the 'grading' of open datasheets measurable04:45
wolfspraulso it's not a matter of a marketing department deciding to declare the company's datasheets 'open'04:45
wolfspraulor the matter of a fan of a particular company or technology to declare the datasheets of that company to be 'the best'04:45
wolfspraulinstead, you try to break it down to provable little factlets04:46
wolfspraulin the end it's all about electrons and physical phenomena04:46
wpwrakyeah, i think it's usually quite black/white. there are not so many corner cases where it's not so clear whether they're open.04:47
wpwrakone set of corner cases would be simply bad documentation, where you can't decode what it's actually meant to say04:47
wpwrakanother set of corner cases is detail information that is willfully suppressed, even though the rest of the manual is open. one example would be the TI CC2591. there, they're very tight-lipped about the transceiver side of the RF interface: http://focus.ti.com/lit/ds/symlink/cc2591.pdf04:49
wpwrakin particular, they don't mention the impedance. the reason is that they don't want you to use this RF frontend with transceivers from other companies (they say so somewhere in a support forum)04:50
wpwrakthis omission isn't easy to spot. and you can in fact find some clue as to what the impedance is, but it's all messy04:51
wolfspraulI can imagine. once you move to actually use a chip, 100 surprises await you that you would have no chance spotting from just reading docs beforehand.04:53
wolfspraulonce you have software running in a chip, the degree to which it is documented is harder to assess04:54
wolfspraulyou can't just look at pins then, and make sure the electronic parameters are fully documented04:54
wpwrakyeah. even if software just talks to the chip04:56
wpwrakand then there are of course those who are just a little terse. they probably don't mean to hide something, they just don't like writing04:57
wpwrakthat's when R&D turns into trial and error :)04:57
wolfspraulno I think you underestimate the degree to which the documentation level is part of a carefully orchestrated business plan and model04:58
wolfspraulmanagers are not that stupid04:58
wolfspraulchip companies always have the problem how they get the new features of their chips through the product into the hands of the actual user04:58
wolfspraulbecause they are only making chips, after all...04:58
wolfspraulso if it's something NEW, how do they teach their customers (who buy the chips) how to use that new thing? how to bring it out to the user?04:59
wolfspraulso _if_ a datasheet is not open, that's because there is someone else on the outside who is doing that on a proprietary IP basis04:59
wolfspraulit's not an oversight, or a lazy or not so lazy engineer deciding how much he 'likes' to write05:00
wolfspraulI don't believe that. It's typically a very clear strategy on how to go from chip feature to end user product.05:00
wolfsprauland since there are such few copyleft hardware hackers like us, they cannot rely on us to help us with that problem, understandably05:00
wolfsprauleven though one would think that open source is the perfect way to get a chip's feature actually adopted05:01
wolfspraulbut in reality it's not...05:01
wolfspraulit's more important to work with a short-list of strategic customers behind closed doors05:01
wpwraksure, but i mean something else. where they're really just tight-lipped. e.g., some control signals where you can probably guess the semantics but they're not properly specified. or where some borderline behaviour is not explained. such as "what happens if you set the enable and the (separate) disable signal at the same time"05:01
wpwrakwhat i mean falls more in the category of stupid little details that add one prototype/test run to your process because you can't be quite sure from the data sheet05:03
wpwrakbut yes, you also have the "fully closed" ones, too05:03
wolfspraulso I think for the parts where you do find high quality open datasheets, that's because that is the quickest way to get the new features/performance of those chips adopted in real products05:05
wolfsprauland conversely, the ones that are not open, that's where the quickest way to get new features into products is via proprietary IP partners who get exclusive behind-the-scenes access anyway05:06
wolfsprauland that is typically a short-list of 5-10 companies05:06
wpwraki think a lot is also just company culture. some almost always write great documentation. plus they care about background material. they expect their customers to be engineers who understand such things.05:06
wolfspraulwhen it has to be more (say because of customer fragmentation), the open strategy will be faster (=work better)05:06
wpwrakand then there are the other who expect their customers to just copy the reference design and hurry on to the next project, without wasting just one brain cell on what the thing actually does :)05:07
wpwrakand yes, those who think you'll need massive hand-holding anyway :)05:08
wpwrakof course, the latter also have quite a cultural dissonance with us. we fully expect to be capable of understanding things well enough to make sense of them, no matter how hard this is :)05:13
kristianpaulwolfspraul: i contacted the guy from the gnss-sdr.ru project, i hoep he join this channel.. as is working ona  fpga + arm device integrated with osgps and namuru !05:19
kristianpaulwell, at least he said that :)05:19
wpwraktuxbrain: you were asking about when my scripted cad exploration/evaluation would be "finished". i did a bit of experimenting with extrusion, but finally realized that the simple algorithm i had in mind isn't sufficient. so i won't have something that is suitable for a fair side-by-side comparison soon. it can make pretty shapes, though :) my extruder lives here: http://projects.qi-hardware.com/index.php/p/cae-tools/source/tree/master/05:20
wpwrakptrude/05:20
wpwrakhttp://projects.qi-hardware.com/index.php/p/cae-tools/source/tree/master/ptrude05:20
wolfspraulkristianpaul: oh nice!05:21
wolfspraulwow that's fantastic05:21
wolfspraulwell let's see whether/when he shows up and we'll give him a warm welcome05:21
kristianpaul:-)05:22
wolfspraulhopefully he doesn't mind openly sharing all his hardware work and discoveries...05:22
kristianpaulyeah, i hope the same 05:22
kristianpaulhis blog is very open, software, schmatics.. board05:22
kristianpaulnot bad !05:22
wolfspraulwhat EDA tool does he use? what's the license of the schematics?05:23
kristianpaulkicad05:23
wolfspraulcool!05:23
wolfspraulit gets better :-)05:23
kristianpaulactually he will send me a bare PCB,05:24
kristianpauli already ordered some samples from maxim.. just to see05:24
kristianpaulof course it doesnt mean i will drop SiGE and my current work05:25
kristianpauli cant find license afaik.. at least for PCB05:26
kristianpauli'll ask him next time he reply 05:27
wolfsprauldo the steps that get you to working results the quickest05:27
wolfspraulwhether that's on m1/sige, or the new arm/fpga/maxim combo05:27
wolfspraulthere is nothing more motivating than getting something to work, and it helps focusing too05:28
kristianpaulsure :)05:28
wolfspraulI forgot why we went the sige route back in the day, it's already so long ago :-)05:28
wolfspraulbut that was specfically after a comparison with maxim, by someone who had used both05:28
wolfspraulevery chip has pros and cons, and people run into different difficulties, also depending on their own background/strengths/weaknesses05:29
wolfspraulbottom line: find the shortest path to working anything. I'm not worried about 'dropping' sige or m105:29
wolfspraulI personally don't find the arm+fpga combo very attractive, I think the all-fpga + softcore approach is more promising. but i could be wrong :-)05:30
wolfspraulI believe there are some chips now that combine arm+fpga in one package, maybe even one die?05:31
wpwrakwhy not the gnuradio approach ? collect the data on the M1 (M1 for interfacing), then send it to the PC for analyzing05:31
kristianpauldunno arm at all, but also ppc i remenber05:31
wolfspraulis there even a problem with analyzing on the m1? I don't know where kristianpaul stands now05:32
wolfspraulit's so slow that I cannot measure progress :-)05:32
wolfspraulneed a more precise measuring tool I think :-)05:32
wpwrakbut as i understand where things are, the problem is still to receive the bitstream into memory, no ?05:32
wolfspraulkristianpaul: I'm just making jokes on you, hope you don't mind :-)05:33
kristianpaulno05:33
wpwrakwolfspraul: yeah, i kinda wonder if he isn't just going around in circles :)05:33
kristianpaulyeah, i kwno your jokes already05:33
kristianpaulno either, 05:33
wolfspraulbut seriously, finding that gnss-sdr guy and trying to team up with him is awesome05:33
kristianpaulthe first aprouch about offline processing was bad idea, but anyway..05:34
kristianpaulwhen you later discover the software you we're using havnnt suport for complex data, oh well..05:34
wpwrakkristianpaul: so you now can receive the bits from the RF chip and store them in memory ?05:34
kristianpaulyeah, but have no sense right now05:34
kristianpaulas the idea is make namuru to work with osgps05:35
wpwrakbut you're think the bits you store in memory are actually the bits the chip sent ?05:35
wpwraks/you're/you/05:35
kristianpaulyes05:35
wpwrakhave you tried to generate a test pattern to verify this ?05:36
kristianpaulin simulation yes i did05:36
kristianpaulnot in real hw afaik05:36
wpwrakhmm :)05:36
kristianpaulhe :)05:36
wpwrakmight be worth a try05:36
wpwraknot being able to make sense of the bits you receive would be quite consistent with them just being incorrect :)05:37
kristianpaulonce i got namuru accumulator to work, sure05:37
kristianpauli dunno why is not wokring well right now..05:37
kristianpaulyeah, sure, i can do that, i have sie board, so it can help to generate a know pattern and compare later in m105:37
wpwrakfor the pattern, what i mean is that you could send, say, the values of an 8 or 16 bit counter. that's something you can verify manually with the scope. then feed it into the receiver and see what bit pattern comes out05:38
wpwrakif it's correct, you know that this much seems to work. you still have the issue that you're synchronous, though, so maybe generate the pattern with a ben :)05:39
wpwrakah, or with the SIE. even better :)05:39
kristianpaulah yes 05:40
kristianpaulwpwrak:  yes yes05:40
kristianpauli edid that with scope05:40
kristianpaulfor clk, and ata signal05:41
kristianpauli just was testing a more complex pattern05:41
kristianpaullike a knwo series of 4 nibbles or soemthing05:41
wpwrakhow wide is you register interface to the receiver in the FPGA ? 8 bits ?05:42
kristianpaul8 bits05:42
kristianpaulbut thats for the acquisition core05:42
kristianpaulright now two bits no more i need for namuru,sign and mag no more05:43
kristianpaulclk, sure :)05:43
wpwrakah, so you're not working on the acquisition core anymore ?05:43
kristianpauli already said, i stoped offline processing05:43
kristianpauli dont wanted to spend more time on it... because osgps will never correlate complex data05:44
wpwrakhmm. why ? :) that just seems to make things harder. any processing algorithm should be just as happy with a recorded bitstream as with a "live" bitstream05:44
wpwraki mean05:45
wpwrakthere's no feedback, so ...05:45
kristianpaulafaik the SoftGNSS code neither,, well i got a update version from fabrizzio but dint looked at it yet or planing soon05:45
kristianpaulsure sure, but i wanted to speed up05:45
kristianpauli took me, long time to learn m1 internals (wishbone, verilog conding, HDl, RTl...)05:45
kristianpaulthen some C, 05:46
kristianpaulthen retake gps theory and try to understand..05:46
wpwraksure :) but it seems that you just made the problem more difficult, by adding a throughput constraint05:46
kristianpaulthen realize my previos mistakes05:46
kristianpaulthroughput constraint?05:46
wpwrakwell, or replacing an "easy" throughput constraint with a harder one05:46
kristianpauldont get it..05:47
wpwrak(throughput constraint) that you have to be able to process the data in real time05:47
kristianpaulah you mean because i shifted the data?05:47
kristianpaulah, yes but when i was told of namuru i had very empties about gnss and dsp05:47
kristianpaulnow i dont, well no soo much :)05:47
wolfspraulqwebirc83241: hi there05:48
wpwraki don't understand how namuru vs. ogpsd would affect the choice between offline/real time05:48
wpwraki picture the architecture of any such thingy as follows:05:49
kristianpaulqwebirc83241: hi :)05:49
wpwrak1) there's some interface that picks up the data05:49
kristianpaulArtyom perhaps? :-)05:49
wpwrak2) there's some digital filtering and such05:49
wpwrak3) there's correlation and such05:49
wpwrak4) there's calculation of the position05:50
wpwrak5) there's some form of output05:50
wpwrakmaybe i skipped a few intermediate steps05:50
wpwrakso it would seem that the difference between real-time processing and offline processing would be mainly in step 1 and maybe step 505:51
kristianpaulit depends 05:51
wpwrakis this correct ?05:51
kristianpaulyes05:51
kristianpaulwell,05:51
kristianpaulyou can do soft correlation (all osgps )or hard correlation (namuru + osgps)05:51
wpwrakso you're porting the baseband processor of namuru to the M1 ?05:53
kristianpaulit is ported, now i dealing with things dint work as spected :)05:53
wpwraki see. okay, that explains why you're still fighting with clock domains and such :)05:54
kristianpaulwell, that already was managed05:55
wpwrakhow do you know before you see it work ? ;)05:55
kristianpaulhehe dont make mee doubt again05:55
kristianpaul:)05:55
wpwrak*grin*05:56
kristianpaulhe, is okay, at least this time i lear from you and did a small test program that read and write a  namuru control register 05:57
wpwrakthat's or course one of the problems here - very few intermediate points where you can check things. what might help is a reference bitstream from an RF frontend (or some synthesized bitstream). if you had that, you could feed it into your system and compare the results.05:57
kristianpaulbut afaik spartan6 popup some other issue with routing clocks, and i have to asume that for now and do a workaround05:58
wpwraki wonder if the namuru guys don't have such a reference data stream. it would make testing a lot easier than just trying to use live signals every time05:58
kristianpauli could ask, Peter Mumford sounded friendly last time i asked to him.05:59
wpwrakmaybe you could just ask how they debug all their stuff. maybe all they do is really just plug it an antenna and see what happens. but maybe they have some test data sets with know inputs and known outputs.06:03
kristianpaulatenna, yes thats the most likelly i think06:04
kristianpaulah, yes there was a ugly truought limitation with mm1, for the ethernet part,06:06
kristianpauleven in complex mode, too much data to transmit and no enogutht bandwicth.. and of course no way to decimate 06:07
wpwrakhmm, how much data does the sige generate ? 8 Mbps ?06:08
wpwrakand how long do you need to receive before the algorithm can "see" something ? 30 seconds ? more ?06:09
wolfspraulI think it can theoretically be many minutes, curious what kristianpaul says...06:09
wolfspraulI believe he has already downloaded some start/static data to help the algorithm speed up, that was my understanding at least. maybe incomplete or wrong understanding.06:10
wpwrakwolfspraul: you mean the almanac ?06:10
wolfspraulyes, maybe that, not sure06:11
kristianpaulwpwrak: in complex mode 2.048MSPS (4 bit sample), real mode 16.384MSPS (2bit sample)06:11
kristianpaulwpwrak: remenber gps telemtry is 50bps06:12
kristianpaulsure you can cheat and get a fresh almanac and load to osgps, also give a warm position, so things can speed up06:13
wpwraks/cheat/optimize/ ;-)06:14
kristianpaulsure,06:14
kristianpauland yes, every 30seconds you get a frame if remenber well06:15
wpwrakhave you successfully decoded an entire navigation message yet ?06:15
kristianpaulNO06:15
kristianpaul:-)06:15
wpwrakah, no need for an almanac then :)06:15
kristianpaulno yet06:15
kristianpauli confir signal, but wasnt able to track it..06:16
wpwrak30 s per frame .. so that's a minute to be sure you get at least one, no matter when you start (in relation to the atomic half-second)06:16
kristianpaulof course at leat in 12 seconds or less no 6, you can get a TLM06:17
wpwrakwith real data, that's 4 MBytes * 60 = 256 MB. hmm. can M1 store to uSD ?06:17
kristianpaulNO06:17
kristianpaul:(06:17
wpwrakwith complex data it would only be 60 MB ...06:17
kristianpauli created a ramdisk in rtems for data06:18
kristianpaulbut afaik all samples i have are complex06:18
wpwrakand you can't arithmetically convert complex to real ?06:18
kristianpauli dont know how but in therory yes,06:19
kristianpaulbut is not worth06:19
wpwraklike in  r = hypot(q, i);06:19
kristianpaulevery complex sample is 4x a real one06:19
kristianpauli undertand06:19
wpwrakwell, you would have a lot less data, it seems06:19
wpwrak(TLM in 6-12 seconds) but only if you're synchronized with the atomic clock, right ? otherwise you still need to receive for maybe 40 seconds, correct ? (i.e., if you begin receiving just after the first bit of a TLM, then you need to keep receiving until the next message, plus the time until the end of that message's TLM.)06:23
wpwrakso just looking for TLMs only helps if you want to make many acquisitions and try your luck. but not if you want to be sure that the data you're looking for must be the acquisition.06:24
kristianpaulevery subframe (6s each) have a 8 bits preable so you can catch that and sync the rest from there06:24
wpwrakaah, i see06:24
wpwrakand you've received and decoded subframes yet ?06:25
kristianpaulno, as i said just detected signal, 06:25
kristianpaulbut i'm working on it06:25
wpwrak(TLM) ah, i see that there's more than one in a message. that helps.06:26
kristianpaulonce the accumlator works, i can do a loop that will slew the code nco until get a interesting treshold06:27
kristianpaulof course this loop isnt aware of dopler.. and a moving satellit,06:28
kristianpaulso thats whena  "pull-in" algorthm saves the day :)06:28
wpwrakpick a satellite on a tangential course ;-)06:28
kristianpaulsure i can06:28
kristianpaulgpredict tell me06:29
wpwrakor maybe one that's stopped for maintenance :-))06:29
kristianpaullol06:29
kristianpaulyou need  that subframe before now satelly health..06:30
kristianpauls/now/know06:30
kristianpaulthat could be donne in india,i think their gnss implementation dont move so much...06:32
wpwrak(gpredict) nice :)06:33
kristianpaulgpredict plus a fucunbe dongle is great, you can get weather telemtry, listen ISS.. and what not06:33
wpwrakaah, that's what you want the funcube for !06:37
kristianpaulyeah, i need an updated forecast :)06:38
kristianpaulalso listen some cubesatswith open telemetry06:39
kristianpauleven a coming artist related sat, (sat.mur.at)06:39
kristianpauland my prefered FM station ;)06:40
wpwrakhmm, no GPS satellites over the south pole. that's about the worst place to have your GPS fail :)06:53
kristianpaulhah06:56
wpwrakhmm, gpredict seems to lack a nice trajectory view. e.g., something like the trajectory for the next hour06:59
wpwrakand it's hard to get rid of the ground area06:59
kristianpaulyou can modify time 07:15
kristianpaulbuy yeah, there are some shortcomings for prediction..07:16
wpwraktuxbrain: (cad) that is ... maybe i do have a solution for the extrusion problem that's not too hard to implement ... hacking ...07:32
Action: kristianpaul zz07:56
wpwrakkristianpaul: caffeine !07:59
kristianpaulwpwrak: tea dint worked well...08:04
ArtyomIs anyone here? ;)08:04
kristianpaulhi Artyom 08:04
kristianpaulnice to see you around :)08:04
Artyomhi KristianPaul :)08:04
Artyomthank you. I've seen your discussion here couple of hours ago - very interesting ;)08:05
kristianpaulah,yes wpwrak always made good questions :)08:06
ArtyomI face very similar problems with debugging hardware...08:06
ignatius_Hi, all. Finally solved the boot up issue. I figured out why. Apparently, I needed to wipe the NAND clean and reinstall everything. Which I was trying to avoid. Oh well. No, I have a new problem; the rootfs doesn't use the entire NAND space... there are about 4 seperate "tmpfs" entries, each using about 40MiBs.. So, my question is, how do I get the system to use all of the space in the rootfs?08:06
kristianpaulbtw, before i fogot what's the license of gps_for_www.zip ?08:06
kristianpaulArtyom: how you sort it out?08:07
ArtyomGood questions. My aim was to make a free project in order to attract people to it. I didn't put any information about license because I don't know much about it. May be you can suggest something?08:08
kristianpauldual lincense GLPv3 and CC-BY-ShareAlike version 3.0 :)08:10
ArtyomAnd how it can be done? I only have to add a text file with license to the source?08:12
kristianpaulyeah, a file called LICENSE should be okay for thre PCB design i think..08:13
Artyomok, I will add such a file. 08:16
wpwrakLICENSE or COPYING are common names. sometimes people also put licensing information into a README. here's an example from one of my projects: http://projects.qi-hardware.com/index.php/p/ben-wpan/source/tree/master/COPYING08:17
wpwrak(this one has GPL v2+, LGPL v2.1+ and CC-BY-SA 3.0)08:17
Artyomthanks for this example :)08:18
kristianpaulso many questions, also i'm curious about your work around osgps, and the fpga-arm board..08:18
wpwrakArtyom: regarding testing, so how do you test/debug ? just plug in an antenna and start receiving ? or do you have some reference data that you've recorded/generated and that you can replay ?08:19
ArtyomWith namuru I wrote some hdl-test at first. And I fed signal from the file in this test. This way I checked the acquisition process.08:21
ArtyomBut it is very slow08:21
ArtyomFor traking I tried to work with real signal from antenna08:21
ArtyomBut with no success for now08:22
wpwrakthe test with the file would be in a simulator ?08:22
ArtyomBut I used my own code for tracking wich I checked in matlab/scilab with Akos code08:22
ArtyomTest file was made with my hardware (gnss-sdr front-end)08:23
ArtyomAnd I processed this file with matlab/scilab code to find out which satellite are in view (and to know doppler)08:24
ArtyomBut 08:24
Artyomtracking doesn't work now08:24
ArtyomSorry, I have to leave now. I will join this chunnel as soon as I will come back to my pc. I will definitly try to answer on all questions.08:25
qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-07302011-0543/09:01
wolfspraulzedstar: did sujan publish any blog post about the nepal trip?12:26
kristianpauloh, i dint knew tha tracking dint worked...14:06
kristianpaulor he meant that for the simulated data?14:06
kristianpaulwolfspraul: nice choice from ME cartoon :-)14:09
wolfspraulyes14:09
wolfspraultomorrow is 08-0114:10
wolfspraulI am so totally behind again14:10
wolfspraulcleanup more tomorrow...14:10
kristianpaulok. i'll try to cleanup what i wrote there then 14:13
wolfspraulhe :-) thanks!14:16
wolfspraulbut no worries I'll go through...14:16
kristianpaulgreat, no worried now then :)14:24
wpwrakwolfspraul: hmm, no word about atben/atusb production finishing and them being available now ? june only had the pcbs, no smt/testing/shop yet14:33
wpwrakmilkymist news: rc3 boards have been made and are being tested14:42
wpwraknanonote: maybe my test point map ? http://lists.en.qi-hardware.com/pipermail/discussion/2011-June/008140.html14:44
wpwrakmilkymist: openwrt !!!14:45
wpwrakmilkymist: maybe also point to all the "paperwork" - leaflet, flyer, box ?14:47
kristianpaulwpwrak: please edit the wiki14:47
kristianpaulopenwrt is nice to quoute indeed14:47
wpwrak(edit) naw, i think it's better for consistency if things pass through the editor. if i just dump things there, it's probably harder to find them. also, it may be more awkward to remove items considered irrelevant14:51
kristianpaulhhaha, come on14:52
kristianpauldump it here http://en.qi-hardware.com/w/index.php?title=Copyleft_Hardware_News_2011-08-0114:52
kristianpaulremove is prety easy, so cares about it :)14:53
wpwrakqi-hw general: maybe that CERN is now also officially interested in open hw ? http://lists.en.qi-hardware.com/pipermail/discussion/2011-July/008409.html14:53
kristianpaulno cares*14:53
kristianpaulyes it seems14:53
kristianpaulhe even sabastien http://www.ohwr.org/projects/tdc-core14:55
wpwrakkristianpaul: removal is tricky. you're undoing work someone else has spent time on. sometimes worse, you may not like where an emphasis is put but it may be hard to change that.14:55
kristianpaulok no more excuses,14:55
wpwraknanonote: maybe also the navigation board, for which volunteers are wanted ? http://lists.en.qi-hardware.com/pipermail/discussion/2011-June/008344.html14:59
kristianpaultuxbrain: also offered some stuff for people interested to work in arduino support for atben15:00
kristianpauldunno if he want such us publicity?15:00
wpwraki don't remember seeing that15:01
kristianpaul9382:257489:16:33 < tuxbrain> talking about arduino modules: who was here interested/skilled to develop an Arduino version of the atben/atusb? also DocScrutinizer you were almost skilled in rf , how about bost mA of emision on the atben/atusb?15:04
wpwrakah, that may have been a reference to [g2] whom i mentioned to tuxbrain earlier. [g2] was planning to make an arduino with built-in ieee 802.15.4. haven't heard from him for a while, though.15:06
rohwolfspraul: if you want you could do some pre-marketing for the mm1 batch with this photo http://yamato.hyte.de/tmp/logotest/SANY0029.jpg15:06
rohsomething like 'will feature the new logo' or so15:07
zedstarwolfspraul: hi....he came back a couple of days ago so will chase all that up....out there it was hard to do anything internet related15:07
rohwpwrak: arent there already such boards?15:12
rohhttp://www.logos-electro.com/zigduino/15:12
wpwrakroh: we wanted to make a really small one15:15
wpwraks/we/he/15:15
rohwpwrak: ah. so like a smd-jeenode with an atrf?15:31
rohand pcb antenna?15:32
wpwrakdunno jeenode. i think the one he had in mind is some nano-arduino. dunno about the antenna15:32
kristianpaulbtw anyone know a code snipet to convert and integer to  char bar?15:33
rohkristianpaul: i used something like that.. on arduino.. mompls15:34
rohLcdBarGraph.cpp15:36
kristianpaulcan you pastebin it somwhere?15:37
wpwrakkristianpaul: while (n--) putchar('#'); putchar('\n');   ? :)15:39
rohhttp://pastebin.com/MZJGB8iB http://pastebin.com/3K7e2t3K15:39
rohfound it on the web somewhere and patched it to allow multible bars in multiple lines and starting locations15:40
roh-LcdBarGraph::LcdBarGraph(LiquidCrystal* lcd, byte numCols)15:40
roh+LcdBarGraph::LcdBarGraph(LiquidCrystal* lcd, byte numCols, byte startCol, byte startRow)15:40
kristianpaulhum...15:40
rohit draws the bars by using the lcd lib and these predefined special chars15:42
kristianpaulnot wat i was looking for, for i ditn knew it of this putchar() xD15:42
wpwrakgrmbl. meshlab is really trying my patience ...16:04
kristianpaulokay the carrier NCO is getting stuck.. why?..16:17
wolfspraulroh: thanks for the heads up about picture! yes, that's the nicest of the 4, you are right. Will mention it.16:19
Action: Jay7 thought about CF-6lowpan16:56
Jay7may be useful for CF-only devices16:56
Jay7e.g. for zauruses which have no wireless ifaces (except SL-6000l/w)16:57
wpwrakJay7: yes, you could make that. why not. CF isn't very popular these days anymore, so it's probably not something you could sell much, but you may make friends in the zaurus community :)20:20
kristianpaulhi Artyom 21:03
Artyomhi Kristieanpaul :)21:04
kristianpaul16:04 here21:04
ArtyomSorry that I have to leave very soon last time21:04
kristianpaulnp, actually i got sleep at that time :)21:05
ArtyomI read that you faced similar problems with debugging osgps+namuru21:06
kristianpaulwell, currently  the carrier nco is not working..21:06
kristianpaulbut yes with osgps itseld i wasnt able to track signal21:06
ArtyomIn my case I think that the problem is in interface between arm7 and fpga.21:07
kristianpaulwich fpga are you using?21:07
ArtyomThay are connected through a async static memory interface21:08
kristianpauloh21:08
ArtyomI use spartan 3e and lpc247821:08
Artyomthis board: http://starterkit.ru/html/index.php?name=shop&op=view&id=10 (the page is in russian :( )21:09
kristianpaulhe, i have a SIM508 around :)21:10
kristianpaulnad this board http://en.qi-hardware.com/wiki/SIE21:11
kristianpaulbur currently i'm working all in the FPGA21:11
kristianpaulbut this async interface still preserver namuru memory map?21:12
Artyomno, I had to change memory map because of memory addressing...21:14
kristianpaulwhat about in osgps?21:14
ArtyomI tried to use only single chanel in FPGA.21:14
kristianpaulsame here :)21:14
Artyomabout osgps: i slowly studing the code. I made it work with file that I wrote with my front-end. And I received a PVT-solution21:16
ArtyomNow I want to split osgps in two parts. One (the correlator part) will work on pc and the second part will work on ARM. This way I want to debug ARM code and then I will focus on FPGA-correlator21:17
kristianpaulis not that similar to what is gpl-gps?21:18
kristianpaulwell, i dint check that code but i read the thesis21:18
kristianpaulwow, PC <-> arm <-> fpga, sounds line a lot of fun :-)21:19
kristianpaulalas, will be nice have a portable gps receive, dont you think?21:20
kristianpaulor what is you final aplication?21:21
ArtyomActually at first I wanted to use gpl-gps+namuru. But later Iater I switched to gpl-gps because it continued to develope and because there is a full-software solution.21:21
Artyommy final aim is a portable receiver :)21:22
ArtyomBTW have you seen this project: http://code.google.com/p/cu-hw-gps/21:23
kristianpaulnope21:23
Action: kristianpaul click21:23
kristianpaullooks itneresting, but no activity at all?21:24
kristianpaulah,, nv there are commits from 200921:24
Artyomyes21:25
ArtyomBut interesting pdf connected with this work: http://instruct1.cit.cornell.edu/courses/eceprojectsland/STUDENTPROJ/2009to2010/ams348/hw_gps_receiver.pdf21:25
kristianpaulis this gps work fo you related to some University, or something on your own?21:28
kristianpaulthe lately aboyt fpga+arm21:28
ArtyomI would say both. There are some potential applications in the university (some works connected with pseudolites). But everything have started as my own initiative...21:30
kristianpaulnice :-)21:30
kristianpauland you have a repository or something to follow your commits?21:31
Artyomno, right now only some source on my pc. No time to spend on studing svn or git :(21:32
kristianpaulhe, sure :)21:32
ArtyomAnd what is your aim? ;)21:33
kristianpaulportable receiver as well21:34
kristianpauli'm currently using milkymist board as i told you in last mail i think21:35
kristianpaulmilkymist one*21:35
kristianpaulis not that portable but the idea once work is make it portable  of course :)21:36
kristianpaulor may be a combo with a mips cpu21:36
kristianpaulbut first it need to work :)21:36
kristianpaulbtw how do you replaced the lpm_counters from the time_base core?21:38
kristianpaulwith another xilinx library or own hdl code?21:38
Artyomhdl-code21:38
kristianpaulthis is odd, tic_enable and pre_tic_enable seems working21:39
Artyomfor all altera-specific functions21:39
kristianpaulbut my carrier NCO is stuck on same value 1628210621:39
Artyomand how do use osgps on milkmist? Do you have a soft-cpu on spartan?21:39
kristianpaulyes21:40
kristianpaullm3221:40
Artyomhow do you check it (NCO)?21:40
kristianpaullattice mico32, a BSD-like licensend CPU21:40
kristianpauli was reading CODE_MEASUREMENT register21:40
kristianpaulso i do same with CARRIER21:41
kristianpaulso now i'm checking 10 msb bit from acum_reg register in the scope 21:41
Artyomand did you check carrier-nco output on scope?21:42
kristianpauli'm on that righ now21:42
kristianpaulas reading the register gave me same value...21:42
kristianpaulArtyom: milkymist is a whole SoC http://milkymist.org/socblock.png21:42
kristianpauland yes it can run uclinux and rtems21:43
Artyomlooks very complicated for me ;)21:43
kristianpaul;-)21:44
kristianpaulso as osgps is ansi C will be easy to port, but right now i'm doing some basic tests in software, at least a loop with the code slew and see what i catch21:44
kristianpaulbut that will be after i get carrier nco to work..21:44
kristianpaulthis is code for carrier nco http://paste.debian.net/124712/21:46
kristianpaulmay be you can check my hdl implementation?21:46
kristianpauli jsut lean some verilog so i'm still writing buggy code ;-)21:46
kristianpaulahh sorry, that is not need to port21:48
kristianpaulwas just timebase.. and code gen i think..21:49
ArtyomI'll try but I mainly write in vhdl (and converted namuru to vhdl...). though reading verilog is not a problem.21:50
kristianpaul(vhdl) oh, nice so you should be very good on the topic :) 21:51
Artyom:)21:52
kristianpaulso you re-wrote all namuru to vhdl?21:58
kristianpaulhttps://github.com/kristianpaul/milkymist/blob/gps-sdr-testing/cores/namuru/rtl/namuru_time_base.v#L8221:59
Artyomyes, I decided that it's better to use language that I know a little then to spend time on studying new language...22:01
kristianpaulhow do you debug namuru? just reading control registers and such or?22:02
ArtyomI wrote simple test for each vhdl-file to be sure that it works correctly.22:04
kristianpaulah, testbench i see22:04
Artyomyes22:04
ArtyomAfter that I wrote a simple test for full-corelator. This test read data from a file with GPS-signal-record and put data to namuru input. It also made some initialisation (like setting tic-period and all the rest). Also prn number was chosen and doppler set to correct value. And finaly 22:06
ArtyomI passed through all delays22:06
ArtyomThe result was written to the file. And I made a graph and saw that there was a correlation peak22:07
ArtyomAfter that I started to experiment with hardware22:07
Artyomhow can I show you my code for time_base?22:09
kristianpaulpatebin.com perhaps?22:09
kristianpaulpastebin.com*22:10
kristianpaulor whatever you like :)22:10
ArtyomGood place, I didn't know it before. And I have no other suggestions: http://pastebin.com/VKVzm5Cr22:12
kristianpaulwhy reset value 000000000000001111111111?22:14
kristianpaulahh 255..22:15
kristianpaul1023*22:15
kristianpaulbut i tought as it been a 24 bits counter, it was all 24 bits at HIHG as reset value22:16
ArtyomI don't remember exactly. But there were some trouble with all '1'22:16
kristianpaulmay be i'm facing it ;-)22:16
kristianpauli'll try22:16
Artyommay it was connected only with modeling in Isim...22:20
Artyommay be it was connected only with modeling in Isim...22:20
Artyomif you are interested in vhdl-code I can send it to you22:26
kristianpaulwell, yes it will be nice, also to compare a code i generally understand in verilog to compare vhdl 22:26
kristianpauli will not switch to vhdl at least milkymist project does, too, no anytime soon i think22:27
kristianpaulbut is nice to be open to other coding22:27
Artyommay be I will switch to verilog in future ;)22:29
kristianpaulnice !22:29
kristianpaul:-)22:29
kristianpaulthis PVT-solution i dont understand very well, what is it?22:29
ArtyomPosition Velocity Time ;)22:29
kristianpaulahh !22:30
kristianpaulbtw are you aware of rtklib project http://www.rtklib.com/ ?22:30
ArtyomOh, I've seen it and I would like to play with it in the future. But now I don't know a lot about it. How to use it. And what kind of receiver do I need to play with it22:32
ArtyomThere is also: http://www.gogps-project.org/ As I understand they are similar (or am I wrong? ;) )22:33
ArtyomI think I should leave you for today. Or I would not get up in the morning ;)22:34
kristianpaulhe, sure22:34
kristianpaulthanks for hangong22:35
ArtyomThanks for conversation it was very interesting :)22:35
kristianpaulYou're very wellcome to stay !22:35
kristianpauli mean later, first sleep :)22:35
Artyomthanks :) I will definitly do both ;)22:36
Artyombye22:37
kristianpaulwpwrak: check this out ,doesnt look familiar to you http://starterkit.ru/html/index.php?name=shop&op=view&id=10 ?22:37
kristianpaularghh, carrier still stuck..22:41
Action: wpwrak curses meshlab22:43
wpwrakkristianpaul: no, doesn't look familiar. should it ?22:43
kristianpaulsie a bit22:44
wpwrakhmm. distant relative :)22:44
kristianpaulheh22:45
wpwrakheh, finally found a file format ImageMagick can't handle: STL (3D meshes) :)23:07
kristianpaulhaha23:37
kristianpaullol vhdl code looks nice when vim colored it compared to verilog xD23:49
--- Mon Aug 1 201100:00

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