#qi-hardware IRC log for Sunday, 2012-03-04

DocScrutinizerHAH! http://javad.com/downloads/javadgnss/publications/20112312.pdf01:30
Action: chrisb takes a look01:35
DocScrutinizer(about GPS et al)01:36
DocScrutinizeranother reason why you should ignore tech specs originated in USA, and better do your own technically based specs for anything you want to build01:37
DocScrutinizerif you don't get it what I'm talking about, look at GSM vs the mess they got over in USA with mobile phone bands01:38
DocScrutinizerand they managed to mess it up despite GSM got invented and defined over here in Germany, in a proper decent way01:39
DocScrutinizernow figure where you end with anything invented and specified in USA01:39
DocScrutinizeror simply read that pdf to get an idea01:40
chrisbDocScrutinizer: my, my let's not generalize.01:46
chrisbDocScrutinizer: there is a business case as well as a technical case here01:47
DocScrutinizersure01:47
chrisbDocScrutinizer: are you familiar with LightSquared and its financial backer?01:47
DocScrutinizerbut the business case is always more important than the technical one, when the story starts in USA01:47
DocScrutinizerand no, I never heard of lightsquare before I read that paper some 30min ago01:48
DocScrutinizerbut I'm familiar with quite some of the technical headache that got quoted in that paper01:48
DocScrutinizerregarding GPS performance01:48
DocScrutinizeras well as general mindset in industry regarding how to implement sth like a RF rx filter01:49
DocScrutinizeryou usually simply copy the reference design01:49
DocScrutinizernot validate the technical fundamentals that it's based on, like sidebands that need to either be included or better cut off by such a filter01:50
DocScrutinizerand I know the quality of reference designs originating in USA vs those e.g. done in Germany01:51
DocScrutinizerand the totally differnet mindset that speaks to me from those schematics01:51
chrisbthe US has lots of experience with the difference between bands for signalling position and bands for two way communication02:08
chrisbLightSquared is trying to *innovate* on some unused bandwidth it managed to uncover02:09
chrisbto create two-way communication bands02:09
chrisbhttps://en.wikipedia.org/wiki/Lightsquared02:11
chrisbthey probably have to pump alot of power into a small band, 50MHz, and it bleeds or rings out into well used bands for positioning02:12
chrisbi hope LightSquared can succeed, but not at the expense of lots of GPS receivers now in service02:13
chrisbi'm sure very few of these are built in the US02:13
chrisbit seems to me like a typical issue of innovation in the US02:14
chrisbit is a similar business case to voice-over-IP and video-over-landline02:16
chrisbcable TV and telco fight02:16
chrisbbut innovation mostly wins02:18
kristianpaullol,03:35
kristianpaul(Im not listening! I cant 03:35
kristianpaulhear you!)03:35
kristianpauloh god.. this old school way to work, is a shame..03:37
kristianpaullol, politicians may be doing more noise that the one that the poor filter desing let pass 03:39
DocScrutinizerwell, actually I think lightsquared is a bit... squared to consider using Watts-of-TX BTS in a spectrum next to GPS05:46
DocScrutinizerthe original culprit of all this mess of course is FCC, what else ;-P05:47
DocScrutinizerfor WLAN you have a clear definition of the allowable TX power for each channel, incl crosstalk to adjacent freq. If FCC had considered equally precise specs for GPS band, L^2 never would've thought of such idiotic thing like implementing a terrestrial network some 25MHz next to GPS signals05:50
DocScrutinizeror - depending on that FCC better specs - GPS manufacturers from beginning would've considered using proper filters to get the needed crosstalk resilience into their designs05:52
DocScrutinizerbut since, according to FCC general policy, everybody can build crap as long as it simply fails passively and doesn't actively interfere with any other service, the second assumption might as well be incorrect05:53
ArtyomDocScrutinizer: BTW Have you heard  about very similar situation with GLONASS+GLOBALSTAR like with GPS+lightSquared. 06:07
DocScrutinizernope06:07
DocScrutinizernever heard of globalstar either until now06:08
DocScrutinizerall this brought to my attention by whitequark - thanks for the pointers :-D06:08
ArtyomIt works in L-band also. It's frequencies are aproximately from 1610 to 1625 for uplink. (And GLONASS previously used frquencies up to 1609.4 MHz, but now up to 1605,4MHz)06:16
ArtyomAnd maximum power of user terminal is up to 26 dBm... 06:20
chrisbinteresting06:36
whitequarkDocScrutinizer: fun fact: here, if someone says "smth is of German quality", then s/he means THE quality12:25
whitequarklike for example there are still some metalworking machines in basements of Moscow State University, which originate from 1940's12:27
whitequarkthey still work and are actively used12:27
DocScrutinizeryeah, some time ago we had a record for building that class of stuff12:29
DocScrutinizereven more fun fact: here the almost same term is "Russian tank"12:30
whitequarkrofl12:30
DocScrutinizerit works like a Russian tank12:30
DocScrutinizeror is built like one12:30
DocScrutinizernot very commonly used term though12:31
DocScrutinizeralso usually implies you need a hammer to start the unit12:32
whitequarkvery true indeed12:32
whitequarkthere are a whole lot of Soviet-era tales about different kinds of equipment which won't work unless kicked mercilessly12:34
whitequarksupercomputers, icbms, etc.12:34
whitequarkthere's a thing I miss about that, through. each device came with the full printed schematics, even ones as complex as TV's12:39
whitequarkand you could just get your soldering iron and fix it if it has broken12:39
whitequarkthere are still chunks of solder sold marked as "for domestic use"12:40
whitequark(through I don't really get how you could fix a TV with a soldering iron where tip d=8mm and a cylindrical chunk of solder d=12mm)12:42
rohwhitequark: that was that way in western europe too. something like 30-40 years ago12:47
rohi have repaired radios from then (with tubes) which even had the schems on the backpanel12:48
DocScrutinizererr radio with electron valves, 40 years old?12:52
DocScrutinizerNAH! ;-)12:52
rohDocScrutinizer: yes. sure12:52
rohbut the majority is older. youre right there12:53
DocScrutinizer40 years ago all radios had transistors12:53
DocScrutinizer30 years ago for sure12:53
DocScrutinizernew ones12:54
DocScrutinizerof course there were 40 year old valve radios 40 years ago ;-)12:54
DocScrutinizerbut yeah, that schematics inside the device were a nice surprise whenever you opened one of those12:55
DocScrutinizerwhen you were lucky, they didn't fall apart on unfolding them :-P12:56
DocScrutinizerI think they stopped that habit after several houses burnt down XP12:57
DocScrutinizerthen the schematics came with the user manual12:57
DocScrutinizerbtw a contemporary fridge of a friend of mine came with a user manual with schematics, marked "for Spain only"12:58
DocScrutinizermaybe there it's still mandatory to ship schematics with some classes of appliances?12:59
DocScrutinizeroh sure, on electric stoves you still frequently see schematics13:00
whitequarkDocScrutinizer: burnt down XP?13:18
DocScrutinizerXP == emoticon13:19
DocScrutinizerlike :-P ++13:20
DocScrutinizermaybe better: X-P13:20
DocScrutinizerrather like ;-P * 213:21
Action: roh must admit he started to like pioneer hifi stuff lately. somehow i can find service manuals online as pdf a lot13:25
Action: wpwrak had a valve radio some 35 years ago. mono. a bit bass-heavy but smooth sound if i recall correctly. i think that was the last critter with such technology in our household. i think the B&W tvs already had transistors13:32
whitequarkDocScrutinizer: (gps) http://www.marcus-spectrum.com/Blog/files/gps-0026-lightsquared-spec.gif13:37
whitequarkerr, the accompanying text was in Russian, but approximately this: "a part of L^2's occupied spectrum reaches the L1 band and interferes with M-signal"13:38
whitequarkthe M-signal is IIRC the high-precision military code13:39
whitequarkhm13:39
wpwrakmaking powerful enemies ;-)13:39
DocScrutinizerwhitequark: http://www.colorado.edu/geography/gcraft/notes/gps/gps_f.html  and thanks :-)13:53
DocScrutinizerhttp://www.colorado.edu/geography/gcraft/notes/gps/gps.html#SVSignals actually13:53
DocScrutinizerfrom "  3  GPS Satellite Signals" in menu of first link13:54
DocScrutinizercan't find M-code13:54
DocScrutinizerbut yeah, P-code is "precision code" which probably is not used in regular consumer grade devices13:55
DocScrutinizerfor your convenience: >>The P-Code (Precise) modulates both the L1 and L2 carrier phases. The P-Code is a very long (seven days) 10 MHz PRN code. In the Anti-Spoofing (AS) mode of operation, the P-Code is encrypted into the Y-Code. The encrypted Y-Code requires a classified AS Module for each receiver channel and is for use only by authorized users with cryptographic keys. The P (Y)-Code is the basis for the PPS.<<13:56
DocScrutinizerPRN == Pseudo Random Noise (iirc)13:57
kristianpaullightsquared, sqaured, well, we need more cools applications around L1 Band13:58
kristianpauls/sqaured/squared13:58
kristianpaulI'm not totally sure but you should confirm the lenth of that cryptographic bits to be sure is something worth to be classified :-)14:00
kristianpaullenght*14:00
kristianpaulsure is 7 days... i mean in the navigation message14:02
wpwrakDocScrutinizer: fun problem: assume you want to design a digital probe (for a logic analyzer). it should be high-impedance (>= 1 MOhm), good up to at least 200 MHz, operating at 3.3 V (but larger input range desirable). probe and cable should be as small/thin as possible. also, the thing should be very low-cost. to the point that you can consider it disposable. how would you do it, with industrially available components ?14:59
wpwrakDocScrutinizer: i.e., would you make the probe active ? if yes, how to keep the cable thin ? or passive ? if yes, how to maintain the high impedance ?15:00
wpwrakDocScrutinizer: would you go directly to the LA (maybe 1 m away) or to a bridge board close to the circuit with additional circuit ?15:01
wpwrak(the bridge board could be more complex/expensive. of course, better if it's not needed)15:01
wpwrakas far as "small" is concerned, ideally below 2.5 mm wide/diameter15:02
wpwrakless if possible :)15:03
wpwrakthe simplest would be just a coax, with a FET at its end. then measure the current. but ... how much power does it have to burn ? if i model this as a low-pass RC filter with 10 pF parasitic, it would seem, quite a lot. in the order of 80 Ohm -> 130 mW15:13
DocScrutinizerbridge board, with ~10cm passive wires to DUT15:14
DocScrutinizerand I don't follow your power evaluation, the FET either is closed then n Volt * 0 Ampere, or it's open then 0.0n Volt * ~10mA15:17
DocScrutinizerthe main power dissipation only happens during switching15:17
DocScrutinizerwhich ideally takes no time at all :-)15:18
wpwrakif closed, you'd still need to send a current through it to detect it. and that current would work against parasitic capacitance. 15:18
wpwrakthat would be a single FET, not a pair15:18
DocScrutinizer??15:18
wpwraki.e., open drain15:18
DocScrutinizersure, what else15:18
wpwrakthe pull-up resistor (on the bridge board) would need to pull the signal up quickly enough15:19
DocScrutinizeryeah sure15:19
DocScrutinizeryou usually use current source termination for that15:19
DocScrutinizerI.E. a PNP with a emitter R and basis to a common constant voltage15:21
DocScrutinizervia that common constant vltage you can tune your circuit to give proper rise time with arbitrary parasitic C load (coax cable length) and still not consume more power than needed15:22
wpwrakhmm, that would be considerably more than just a FET at the probe, no ?15:23
DocScrutinizerthe emitter R typically is very low, like maybe 10 Ohm15:23
wpwrakand more than GND plus signal, too. i.e., you'd have at least also a supply voltage15:23
DocScrutinizerthe current source is at your bridge board15:23
wpwrakis there another name for each a circuit ?15:24
DocScrutinizerthe probe has just the FET, maybe a gate pulldown R15:24
wpwraks/each/such/15:24
DocScrutinizeror pullup15:24
wpwraki googled for it and what i found has a current source at the sender and a comparator at the receiver15:24
wpwrakhttp://books.google.com.ar/books?id=oDWRAxCU-g8C&pg=PA326&lpg=PA326&dq=current+source+termination&source=bl&ots=8_HAAvrfZe&sig=a8vj_bzbc6T7HO17FdcLVujcoxA&hl=en&redir_esc=y#v=onepage&q=current source termination&f=false15:24
wpwraksomething like this critter may be nice as a FET. already has gate protection. http://www.vishay.com/docs/66700/si8800ed.pdf15:25
wpwrakmay be a bit slow, though15:26
DocScrutinizerdamn that google link is useless15:26
wpwrak;-)15:26
DocScrutinizer1098 cookies I have to deny, then a page with empty/broken pictures15:27
wpwrakbut you say just 10 cm wire to the bridge board will be fine ? then we can skip the FET trickery :)15:28
DocScrutinizerhttp://de.wikipedia.org/wiki/Konstantstromquelle#Mit_Bipolartransistor15:28
DocScrutinizerwpwrak: I dunno if 10cm are fine or not. Largely depends on properties of your DUT I'd guess15:29
DocScrutinizerwpwrak: reverse http://de.wikipedia.org/w/index.php?title=Datei:Konstantstromquelle.PNG&filetimestamp=20110120191559 for PNP15:30
DocScrutinizerso your FET in probe is from collector of that constant-current source (inner of coax) to GND (shield of coax)15:30
DocScrutinizerR1/R2 probably common for all N probe inputs of your bridge board15:32
DocScrutinizerand make it a poti15:32
DocScrutinizerso you can tune the current like I mentioned above15:32
wpwrak;-)15:32
DocScrutinizerRe is very low, you want the current source be able to deliver some nn mA to a voltage very close to the emitter voltage15:34
DocScrutinizeror you add in a voltage limiter15:34
DocScrutinizersimplest case: a Zener15:34
DocScrutinizerthis way you can define the voltage at FET's source and thus the logic level where it will detect a 115:35
wpwrakhmm, pwm-drive the supply voltage ?15:35
DocScrutinizerof course a simple clamp diode to a constant voltage source common to all probe inputs would be way more versatile and cool15:36
DocScrutinizerthe basic benefit regarding current source of such a clampdiode+CV is that your supply voltage for the current source (at Re/emitter) can be several volts higher than the probe high-level voltage, thus guaranteeing that your current source will properly pull up with full current until high level get reached15:38
wpwrakhmm, that would be for M1. we have a 5 V supply. and signals (to the M1) are 3.3 V15:42
DocScrutinizeryour parameters for the evaluation then are: 1/(200MHz *(2 + 1for_margin)) rise_time; 10 pico As/V parasitic capacity; 3.3V high level; solve for current needed15:42
DocScrutinizercalculate with 100pF max C_parasit, and evaluate the Re for the resulting needed current @ 5 - 3.3 V15:43
DocScrutinizerthen you reduce that current via R1/R215:44
wpwrak1.6 ns rise time. hmm, what kind of FET has that ?15:44
DocScrutinizerHF FET?15:44
wpwrakaah, right. RF FET is the product category :-)15:45
wpwrakkewl, and they go up to 7 A ;-))15:45
wpwrakif i have a FET in the probe, is it safe to have a probe cable > 10 cm. e.g., 50-100 cm ?15:55
DocScrutinizersure, why not17:00
DocScrutinizerall your probes should have same length though, to eliminate propagation delay diffs17:00
wpwrakyeah, it would be tragic to get that bit wrong :)17:03
DocScrutinizeryour signal will move along the coax to the "bridge" at typical cable specific sub-light-speed, and arrive there in the same shape and sequence you fed them into the cable with your FET. Theoretically the cable could be 100s of meters length, and several signal edges traveling along it17:03
wpwrakvery good. now, to find a really thin coax cable ...17:04
DocScrutinizerof course you want to terminate the coax with the right R, to avoid reflections17:05
DocScrutinizersince your current source is basically zero impedance, your termination will be series to cable17:05
DocScrutinizerthis kinda actually defeats the purpose of the current source somewhat, but meh17:07
wpwrak;-)17:08
wpwraki wonder how all those terminations of data lines really work. e.g., to RAM address lines. those inputs don't look like zero impedance to me.17:09
DocScrutinizerassuming your current source is set to deliver less than 50R/3V3 Ampere, it seems still like a good design to me17:10
DocScrutinizerindeed, I'm wondering as well17:11
DocScrutinizererr 3V3/50R of course17:14
wpwrakcool, thanks ! so this doesn't look too hostile. finding a cable may be tricky. digi-key only have thick cables. maybe i should use something like this: http://search.digikey.com/us/en/products/415-0108-250/J10043-ND/175584017:18
wpwrakonly 25 cm, though :-(17:18
wpwrakbut i suppose one could source larger ones of the same kind, if determined17:19
wpwrakeven thinner: http://search.digikey.com/us/en/products/415-0105-250/J10037-ND/175583417:19
DocScrutinizeroh yeah, now it makes sense when my colleague informed me that the Lauterbach debug connector cable is some 200EUR - it's two 16wire flat cables with coax, about as thin as the one you dug up there. ~10cm length, with two nasty plugs, one each end18:20
wpwrakhappyness is .... http://search.digikey.com/us/en/products/U.FL-2LP(V)-04N2-A-(1000)/H11554-ND/165774918:22
wpwrak... and a sharp knife ;-)18:22
wpwrakthat lauterback cable sounds like the kind they make a total of 100 of. worldwide, throughout human history;-)18:23
wpwrakof course, s/happy/happi/18:23
kristianpaul/doxygen/trunk/Config_ext.html18:29
kristianpauloops18:29
kristianpaulhttp://blog.goo.ne.jp/osqzss/e/af7fe0fe97a2c6b886fdb849efcd6c6f?fm=rss18:29
kristianpauldonk ask for details.. as code is no published yet i dont know then..18:32
kristianpaulbut is not a milkymist cpu afaik nios II, also guess al windows.. but is a note proof or working namuru in the wild :)18:33
whitequarkDocScrutinizer: (heatsinks) hm, doubtful19:29
whitequarkI called the module, answered the call on it and hold it for a minute or two19:29
whitequarkit's hardly warm19:29
whitequark*held it. not in the telco sense, just left my phone lying with the call going on19:31
DocScrutinizerwhitequark: you can't really control the OTA situation, so you have no clue how much of its theoretical 100% power the TX actually had to use for your 1min call19:35
DocScrutinizerif you didn't talk, odds are it didn't send much at all (silence TX cancellation)19:35
DocScrutinizerwhen you're close to the BTS, odds are your TX worked with power correction factor like *0.0001 or sth19:36
DocScrutinizerthere'S basically no decent way to test those things, that's why a decent datasheet mentions the heat sink requirements of such a component. They have means to test worst case at that fab where those modules got developed19:37
DocScrutinizersilence TX cancellation probably aka pasue compression or sth like that19:39
DocScrutinizerpause*19:39
DocScrutinizerbasically means that TX doesn't wase power to TX data that's just transporting silence19:39
DocScrutinizerwaste*19:39
whitequarkwell, this datasheet is indeed not "decent"19:44
whitequarkand I indeed did not talk19:44
whitequarkhm19:44
whitequarkwould a microwave oven work?19:45
whitequarkthat is, I put the module in the (turned off) oven which is basically a faraday cage19:45
whitequarkand then open it a bit19:45
viricoh19:45
viricmicrowave19:45
whitequarkviric: wat?19:45
viricI was waiting for the time when microwave ovens appear at #qi-hardware19:45
viric;)19:45
whitequarkwhat's so special about them?19:45
viricthey can kill19:45
whitequarkindeed19:46
whitequarkalmost anything can kill, if used properly19:46
viricnicer, if killing with open hardware19:46
whitequarkwell, you indeed need to keep it open to kill anyone...19:47
viric:)19:47
viricsorry for the pause in your conversation. go on go on19:47
whitequarknah, I said whatever I wanted19:48
viricIsn't there any kernel patch that can make the kernel remember the backlog buffer of every virtual console, regardless of switching'19:48
viric?19:48
viricis this so high tech?19:48
whitequarkoh, just go to lkml and troll them a bit19:48
whitequarkfaster than just asking19:48
virichm could work19:48
viricI've a 80x25 console, switched the vc, and I lost my valuable 8KiB of backlog... what is this trying to optimise?!19:49
whitequarkmemory consumption?19:49
whitequarkthat is19:50
viricon a 7MB kernel?19:50
whitequarkthe consumption of users' memory by this behavior19:50
virichaha19:50
viricmodprobe pcspkr takes more memory19:50
whitequarksigh. DCD is "circuit 109", and DTR is "circuit 108/2"19:51
whitequarkcrazy telcos19:51
whitequarkwhere's circuits 0..99 ?19:51
viric100 is a good start for a three digit number19:51
whitequarkwhy not 000 ?19:52
viricthat may mean writing %03i instead of %i19:52
whitequarkI could argue that DCD and DTR were invented before printf19:52
whitequarkyea, RS232 is of 196919:53
virichistory is full of coincidences19:53
DocScrutinizerwhitequark: yes, a good idea, just don't open it. You won't believe it but microwave ovens are all but tightly sealed for 900/1800MHz19:53
viricDocScrutinizer: the magic space frequency of the grid of holes of the front door19:54
whitequarkDocScrutinizer: er. did I understand you correctly? microwave oven kills 2.4g but allows 900m/1800m to pass?19:54
DocScrutinizeryes19:54
whitequarkhuh19:54
DocScrutinizerlamda/2 seals19:54
whitequarkbut won't lambda/2 seal it for lambda*2 ?19:55
DocScrutinizermaybe they're caled lamda/4 seals, dunno. but yes, they should also work for subharmonics19:55
DocScrutinizerso 2400/2, */4, etc19:56
whitequarkhm, I meant a bit different thing. if my math is any good, then »(2.4g)=.125m, and »(1.8g)=.166m19:56
DocScrutinizerwhatever19:57
whitequarkwhy would holes made to seal .125m pass .166m?19:57
DocScrutinizernot holes19:57
viricthe cage size19:57
DocScrutinizerresonating gap between door and case19:57
whitequarkoh. so it isn't sealed tightly?19:58
DocScrutinizernot at all19:58
whitequark... weird. I'd rather not stand near microwaves anymore19:58
DocScrutinizerhehehe19:58
whitequarkwell, at least it explains why 900m/1.8g escapes19:58
DocScrutinizeryup19:58
whitequarkhm, but why gap and not a close connection?19:59
DocScrutinizerthen place some electric toothbrush or whatever into the oven as well, to make some noise the phone can transmit19:59
DocScrutinizervery first microwave had seals. They were extremely expensive, unconvenient, and easily broke by a simple 0.5mm of random dirt between the elastic metal fibre and the case frame20:00
whitequarkhm, so the sealed approach requires seal to be maintained at all times or the radiation would escape20:03
whitequarkand the gapped one allows for any shit to accumulate in the gap as long as it's precisely the required size20:04
whitequarkcool20:04
whitequarkcan you also explain this... if I'd put a %metal_thing% to an oven and turn it on, I'd get a nice discharge and a small lake of molten metal20:07
whitequarkbut if I put a %metal_cup_for_microwave% into it, I won't see any of the effects. it won't get even slightly hot20:07
whitequarkwhy?20:07
whitequarkI never really understood this20:08
viricthere is a 'metal cup for microwave'?20:09
whitequarkyeah20:11
viricI've never seen one20:11
whitequarkgoogle for "microwave compatible metal [cup]"20:13
whitequarkDocScrutinizer: huh, _very_ interesting20:14
whitequarkyou can run arbitrary code on the CPU of this module20:14
whitequarkwell, "SIMCom Core Software and Embedded Application manage their own RAM areas. Access from20:15
whitequarkone of these programs to anothers RAM area is prohibited and will cause fatal error.20:15
whitequarkbut it's still interesting. I bet they have quite a few holes there20:15
whitequarkyou can also easily downgrade firmware, nothing prevents you from doing that20:15
DocScrutinizerwhitequark: depending on architecture, you hardly will find any hole20:16
whitequarkARM926EJS20:16
whitequarkafaik it's embedded in an ASIC20:16
DocScrutinizeron STE modules we have PSEL and PSEL_SECURE to access peripherals in secure and unsecure mode. Same for RAM access20:17
whitequark"The ebdat4_04Wdtkick function kicks the watch dog."20:18
whitequarkpoor animal :/20:18
DocScrutinizerwe got 3 ARM cores, one for network stack (aka GSM/LTE/*), one for communications between all the subsystems on SoC, and one App core20:18
whitequarkin this case, looks like everything is running on the single core20:18
whitequarkwell, at least that's what I think after looking at a vast amount of docs20:19
DocScrutinizerand even the 4words deep FIFO 5 chan "mailbox" function block has 3 nonsecure and 2 secure channels20:19
DocScrutinizerour customers never get to know about NS and COM core or the functions/processes running there20:20
whitequarkerr, I bet the "display interface" present in this module is made for Nokia 3310 display or equivalent one20:20
GNUtooah baseband....20:21
GNUtoowhat about that:20:21
DocScrutinizerthey might think the whole M7400 modem SoC has just one ARM core to run some general purpose things like AT interpreter and their custom 'apps'20:21
GNUtoohttp://www.pastie.org/private/ana4aetc6o9qzt4w7wtma20:21
whitequarkseriously, it even has some embedded melodies for games20:22
DocScrutinizerpastie.org? no way20:22
whitequarkDocScrutinizer: what's wrong with pastie?20:22
GNUtoocan run on compal_e99 (c155 ) or gta02 or another phone20:22
DocScrutinizertoo much crap20:22
GNUtooah sorry I forgott20:22
GNUtoohttp://paste.debian.net/158551/20:23
DocScrutinizerooh actually it kinda works, at least better than pastebin.com20:23
GNUtooI should implement poweroff I think20:24
DocScrutinizerGNUtoo: what's that?20:24
GNUtoonuttx on the calypso20:24
DocScrutinizerhmm20:25
GNUtooso that you can have an unix-like OS abstraction20:25
GNUtooso you could easily move the upper layer(layer23) on the modem20:26
GNUtoobecause currently layer23 runs on a GNU/Linux computer20:26
whitequarkreminds me of how they embedded busybox in grub220:26
GNUtooso in case of gta02 that prevent the CPU from sleeping 20:26
GNUtooin the case of a feature phone it's even worse20:27
DocScrutinizergrub2 becoming a wanna-be-OS like uBoot now?20:27
GNUtooyou have to connect it to a computer to use it20:27
whitequarkDocScrutinizer: exactly20:29
whitequarkbut it's kinda more advanced. for example, it has a working almost-POSIX shell20:30
kristianpaulnuttx, cool :-)20:30
whitequarkstill very wrong IMO20:30
whitequarkthey could as well compile Linux, bootstrap it from stage2 and then kexec to the desired kernel20:30
DocScrutinizeryup20:31
Action: kristianpaul agree too20:31
whitequarkGNUtoo: DocScrutinizer: maybe you possibly know any AT-command parser libraries?20:32
whitequarkthat is, which could run on a uC with 8k ram20:32
kristianpaulergh, what you want AT?20:32
GNUtoowhitequark, you're involved in basebands too?20:32
DocScrutinizernope, sorry20:32
whitequarksend commands while obeying flow control, decode & encode SMS, parse unsolicited return codes20:32
whitequarkGNUtoo: nope, just a side project which I make for a practical purpose. I use some GSM modules in it and need to interface them20:33
GNUtoook20:33
GNUtooah I get it:20:33
GNUtoo<GSM module> <-----> <microcontroller>20:33
whitequarkyup20:33
GNUtooimagine that you could do <GSM module><-----> peripheral20:34
GNUtoothat would be so cool20:34
GNUtooit would save ton of power 20:34
GNUtooetc...20:34
GNUtoobut for that we need to move layer23 in the microcontroller20:34
whitequarkGNUtoo: I can, actually20:34
GNUtoos/microcontroler/ in the calypso/20:34
whitequarkthey have SDK for that20:34
GNUtoook20:35
whitequarkbut fuck it, I don't want to have anything running on that chip20:35
whitequarker, just look at docs20:35
whitequark"Reset the module frequently is not recommended and it may cause some unexpected fault.20:35
GNUtoowhat's the module model+company?20:35
whitequarkSIMcom SIM900D20:35
GNUtoook20:35
whitequarkit has some interesting features20:35
whitequarklike AT+SIMEI20:35
whitequarkguess what does it do20:35
GNUtoosim I/O ?20:36
GNUtoolet me look in the AT manual20:36
whitequarknot quite. and it's udoc'd20:36
whitequarkAT+SIMEI="<new IMEI>".20:36
GNUtoook20:36
GNUtoolol20:36
whitequarkit "restricts" you by "limiting" the count of imei changes by 220:37
GNUtoolol20:37
whitequarkbut actually, you can "reflash" it by sending 50 proper bytes to a debug port20:37
GNUtoook20:37
whitequarkwhich kinda allows you to change IMEI on what frequency you want20:37
whitequarkeach minute20:37
whitequarkwhatever20:37
GNUtoook20:37
whitequarkhow did they get it certified?!20:37
GNUtoogood question20:37
DocScrutinizerwell, it's probably not part of the cert at all20:38
whitequarkso there's no such point in certification like "make sure the ME could not change its IMEI at the will of user"?20:39
DocScrutinizerbasically cert first of all checks OTA compliance20:39
DocScrutinizerwhitequark: I don't think there's such a point20:39
GNUtooOTA = um ?20:39
DocScrutinizerover the air20:39
GNUtooyes so it's um20:39
whitequark"It is suggested that its better to do some ESD protection in users application to improve the ESD20:40
whitequarkcharacter of the module, especially for the signal connecting to external interface, for example,20:40
whitequarkI like how they write the datasheets20:40
whitequarkDocScrutinizer: huh, I found a (deep buried) appnote which describes how to convert mic-in to line-in20:43
whitequarki.e. what I want20:43
whitequarkit's basically the same thing you have described20:43
whitequarkhm20:46
whitequarkthere's an appnote about how to activate jamming detection20:46
whitequarkis it a common function across GSM modules?..20:47
DocScrutinizernope21:04
whitequarkDocScrutinizer: oh wow. "6.2.24 AT+CENG Switch On or Off Engineering Mode21:25
whitequarkhttp://paste.debian.net/158558/21:26
DocScrutinizerdang, than's nice21:31
DocScrutinizerI wonder if it also supports CID-lock21:32
whitequarkhm. CID-lock?21:33
DocScrutinizerand of course, what the CENG output means, hope they explained it in that manual21:33
DocScrutinizerlock modem to a certain CID21:33
DocScrutinizerso it will ignore C1/C2 and only try to connect to that BTS21:33
whitequarkyes, they did describe it indeed. wait a sec, I'll copy the text21:34
DocScrutinizerCID-lock needed for my very special high accuracy BTS distance based location service21:35
whitequarkhttp://paste.debian.net/158559/21:35
DocScrutinizeras you get TA value only when you associate to the BTS21:35
whitequark6.2.51 AT+CEMNL Set the List of Emergency Number ...................................................15621:37
whitequark6.2.52 AT*CELLLOCK Set the List of ARFCN Which Needs to Be Locked...................15721:37
whitequark(6.2.52, the first line was copied accidentally) this ?21:37
whitequarkah, that's just a channel lock21:40
whitequarkor, hmm, could this still be used to lock to a particular BTS?21:40
DocScrutinizerchannel lock is basically same as CellID lock, yes21:41
DocScrutinizerthough I'm not ure their definition of LOCK is same as mine here21:42
whitequarkit's not really described how this command works, only syntax21:42
whitequarkbut I could try, it's present in my module21:42
DocScrutinizeryou could try to make modem lock to one of the naighbour cells, best not the first strongest one21:43
whitequarkwow, something is happening21:43
whitequarkunknown URC all over my terminal21:43
whitequarkhell yes, it worked21:43
whitequarkhttp://paste.debian.net/158560/21:44
DocScrutinizerso the cell/chan you selected is your servicing cell now?21:44
whitequarklooks like it is21:44
DocScrutinizerhmm yes :-D21:47
DocScrutinizercool, so you could *finally* implement an automated version of http://lists.openmoko.org/pipermail/openmoko-kernel/2008-June/002987.html21:48
DocScrutinizerand thus find out where you are, without GPS21:48
whitequarkerr, it has such a command21:48
DocScrutinizergiven you got an idea about where BTS are21:48
DocScrutinizerI'd probably not try to do this during a cal, though in theory it should work (and call shouldn't stall)21:50
DocScrutinizerdang, is this actually already 4 years in the past?21:54
whitequarkhm, how do I get GK readings?21:55
whitequark*G-K21:55
whitequarkDocScrutinizer: huh, google can locate by cellids22:00
DocScrutinizerthose are special for O2-germany I guess. They are via cellbroadcast SMS channel 221 here22:00
whitequarkand there's a db of Russian cellids22:00
DocScrutinizerwow, so from google we get location of the BTS22:01
whitequarkactually your technique is very well known in Russia. there's even an article in "Hacker" (it's the name) journal, which means that a pretty big amount of people knows about it22:01
whitequarkyeah, google collected cellids too, apart from wifi ones22:02
whitequarkit's quite obvious thing to do I'd say22:02
DocScrutinizernow get TA for 3 or 4 cells (current servicing cell, plus 3 neighbour cells via theis command to select/lock to them) and you got your distance to 4 points on globe22:02
whitequarkopencellid: 250 Russia|Russian Federation 33205 cells22:02
whitequark33k cells, wow22:02
whitequarkheh, very interesting. I wonder how precise can this be22:03
DocScrutinizerwhitequark: location to a circle around your serving cell is very common and widespread. The triangulation with 3 or 4 or 5 cells however probably isn't as it needs that command you found22:04
whitequarkDocScrutinizer: (widespread) I meant the location with a netmonitor tool. there's a lot of stuff on local forums about this22:05
DocScrutinizerhmm22:05
DocScrutinizerwhatever, this module is the first one I get to know of, that can do automatic forced cell reselect22:06
DocScrutinizeralias Cell-lock22:06
DocScrutinizerusually modems offer neither engineering mode and for sure not cell-lock22:07
whitequarkusually modems don't allow you to change imei arbitrarily22:07
DocScrutinizerthat too :-)22:07
whitequarkwell ok, I can source a bag of these if someone wants22:08
whitequarknot a problem, a local supplier will happily sell me any quantity I'd want22:08
whitequarkcomplete with custom fw upgrades service and whatnot22:09
DocScrutinizerwell, you could do some sort of improved-location via triangulation when you monitor the the modem closely for service cell data and catch the values at moment of handover from one cell to another. So you get at least 2 cells with distances to them in that moment, which allows a location to something like a parallelogram with 550m sides22:10
DocScrutinizeratually 2 parallelograms, one for 'left' and one for 'right' side22:10
DocScrutinizerunless you're on the line between those 2 BTS22:11
DocScrutinizerand are lucky enough so sum of TA just matches distance between them22:12
DocScrutinizerin which case you get only one parallelogramoid22:12
DocScrutinizeror actually a section of two circles in that special case22:13
DocScrutinizera very small one22:14
kristianpaulargh, hate race conditions22:15
whitequarkDocScrutinizer: I could also use rssi22:20
whitequarkas in "random signal strength indication"22:20
DocScrutinizer??22:20
whitequarkfor distance measurement22:21
DocScrutinizerdistance from signal strength? muhahaha22:21
whitequarkor, hm, maybe not22:21
whitequarkfor wifi that did not make much sense22:21
DocScrutinizermaybe in an ideal world22:21
DocScrutinizerwithout trees and without buildings22:22
DocScrutinizerand with ideal antenna on your phone22:22
DocScrutinizerand of course without you covering half the perimeter of sight from that antenna22:22
DocScrutinizerlook at the dB values for neighbour cells from the post I linked above. Do they look like steady or correlated to the TA value that *really* gives distance OTA to BTS?22:24
whitequarkok ok I understood it :)22:25
DocScrutinizeresp look at that pair of neighbour cell readouts where I moved just 3 meters22:25
Action: whitequark has imagined this when he read DocScrutinizer's muhahaha: http://en.wikipedia.org/wiki/File:Evillaugh.ogg22:28
DocScrutinizerchan:706, max:-35dB, min_-62dB -- device moved by 60cm(!)22:29
whitequarkDocScrutinizer: btw that trick is confirmed not to work during call22:32
whitequarkit just displays one BTS in the +CENG URC22:32
DocScrutinizerwell, that's just enough, as we're not interested in neighbour cell data once we got the channels. We're just 'roaming' to the next cell by forcing it with that command you found22:33
DocScrutinizerand that shouldn't (but might nevertheless) interrupt the call22:33
whitequarkah. hm, I'll test22:34
whitequarkhm weird, it just stopped displaying neighbour cells22:35
whitequarksimple, I forgot to remove celllock22:36
DocScrutinizerI think I like to get one or two of those nice modules :-D22:36
whitequarkno problems. email your postal address at whitequark@whitequark.org then22:37
DocScrutinizersure thing22:37
whitequarknot sure what method should I use to send them, through22:38
whitequarkfirst-class mail? dhl?22:38
DocScrutinizerhmm, usual letter?22:47
whitequarkthree words: "Russian Postal Service"22:47
whitequarkI bet they have a 10-ton press for these or whatever22:47
whitequarkmaybe they run with the truck over letters22:48
DocScrutinizer:-S22:48
whitequark(apart from the fact that 12g may be a bit too heavy for a "usual letter". and the electronics inside will arouse customs)22:48
DocScrutinizernah, 12g are fine22:48
Action: whitequark never sent these in his entire life22:49
DocScrutinizerusual letter is either 20 or 50g max, can't reacall22:49
DocScrutinizerwell, whatever you think may work best22:49
whitequarkhm ok. I think I'll send them maybe at next Sat22:50
DocScrutinizerno hurries22:50
whitequark... you have some Australian friends, don't you? :)22:50
DocScrutinizerer, kinda22:50
DocScrutinizerRaster comes to mind, but he's probably in Korea most of the time22:51
DocScrutinizerwhy?22:51
kristianpaul12g!! is not heavy IMHO22:52
whitequarkDocScrutinizer: "not worries" is a purely Australian English expression afaik22:52
DocScrutinizerooh, didn't know that22:52
whitequarkthat's what I thought of reading your message22:52
kristianpaulbtw this module you talking about is a modem?22:53
whitequarkkristianpaul: yeah. AT interface, analog I/O22:54
whitequarkit can GPRS, through, and has full TCP/IP stack inside22:54
whitequarkeven with HTTP&FTP22:54
kristianpauldoes it work with the nanonotr22:54
kristianpaulahh too bloated (HTTP&FTP)22:54
kristianpaulok22:54
kristianpaul:)22:54
whitequarker22:54
whitequarkyou can omit using that22:54
whitequarkjust use it as a dumb modem with pppd22:55
kristianpaulyeah, i was thinking in that22:55
whitequarkI'm sure NN has an uart22:55
kristianpaulsoon or later i need som kind of gprs/edge/3G/ for my nanote22:55
kristianpaulsome*22:55
DocScrutinizerno 3G22:55
kristianpauledge?22:55
whitequarkbut! I seriously doubt its battery can provide the required 2A peak@3.3V (actually 2.7V-4.8V)22:55
whitequarkno edge22:55
DocScrutinizerthis is 2.5G aka EDGE22:55
kristianpaulwtf 2A!!22:56
whitequarkthat's GSM22:56
whitequark2A _peak_22:56
kristianpaulyeah, not my field..22:56
whitequarkin idle mode it eats 1.5mA22:56
whitequarkin tx/rx that may be in range of 10mA-200mA avg22:56
kristianpaulhad you tried make it work?22:56
kristianpaulwith the nanonote and pppd22:56
whitequarkI don't have NN22:56
kristianpaulWHAT?!22:56
kristianpaul;-)22:56
DocScrutinizerme neither22:57
whitequarktuxbrain is kinda slow on delivery22:57
kristianpaulI tought, you... all this time22:57
whitequarkwell no22:57
kristianpaulOMG still delivering?22:57
whitequarkyeah, more than 2 months22:57
kristianpauloh well..22:57
whitequarkI mainly want it as a testbed for 6lowpan22:57
whitequarknot that I have much use for NN apart from that22:57
whitequarkI have an Android phone22:57
Action: whitequark hides from DocScrutinizer 22:57
kristianpaulergh, i havbe dumb phone22:57
kristianpaulbut need at least gprs for mutt and irssi when a hurry22:58
whitequarkthis module is basically what's inside your dumbphone, but without keyboard and screen22:58
kristianpaulrss reader will be cool to22:58
kristianpaulsize?22:58
whitequarkseriously, you can just add some buttons and a display, and it'll be nokia 331022:58
kristianpauli have a sim548 too big22:58
whitequark33x33x3mm22:58
kristianpaulbuilt-atenna?22:58
whitequarknope22:58
kristianpaulah22:59
whitequarkbut you can attach any easily if you know how to solder22:59
whitequarksrsly, not much of a problem22:59
kristianpaulyes i supose i do :)22:59
kristianpaulwell, i'll wait for lightsquare satellites to deploy, lol23:02
DocScrutinizeraen't those already flying?23:02
kristianpauli guess so, but enought to cover worldwide?..23:03
DocScrutinizerno way23:03
whitequarkbecause?23:04
kristianpaulwill crash with GPS ones ;) 23:04
kristianpaul(kidding)23:05
whitequarkthat's the privilege of russian sats23:06
whitequarkthere was some funny/horrifying story (depending on whether you are a telco engineer) when one of russian sats malfunctioned and kicked an Iridium sat off the orbit23:08
DocScrutinizerhey yeah, we 'recently' had some collision up there X-P23:09
whitequarkDocScrutinizer: huh? another one?23:09
kristianpauldecay all days :)23:09
DocScrutinizerI only know of one, can't recall details, but the crash generated like 35000 new tiny objects that all need to get monitored by NASA or whoever is responsible for that23:10
whitequarkyep, it's that one23:10
DocScrutinizerthey wait for a chain reaction on next or the crash after that23:11
whitequarkhttp://www.theregister.co.uk/2009/02/12/satellite_collision/23:11
whitequarkthe count of objects is 500-600 through23:11
DocScrutinizerdepends on size where you start to call it an object, I guess23:12
DocScrutinizermaybe the trackable ones are actually only 50023:12
DocScrutinizerthe real headache are those you can't track as they are too small to find them with radar23:13
kristianpaulwhere is(are) that radar(s) btw?23:14
DocScrutinizerNFC, ask NORAD or NASA23:14
kristianpaul;)23:14
DocScrutinizerNORAD probably23:14
DocScrutinizer>>Die Überwachung schließt von Raketen übrig gebliebenen Weltraumschrott sowie das Aufspüren, Validieren und Warnen vor Angriffen auf Nordamerika mit Flugzeugen, Raketen oder Weltraumfahrzeugen ein.<< http://de.wikipedia.org/wiki/North_American_Aerospace_Defense_Command23:16
kristianpaulurgh that sword scares..23:16
whitequarkphallic symbol23:18
DocScrutinizer>> To accomplish the aerospace control mission, NORAD uses a network of satellites, ground-based radar, airborne radar and fighters.  www.norad.mil  About. Aufgerufen am 17. Juni 2010 <<23:20
whitequarkmhm. "airborne radar and fighters"?23:21
DocScrutinizer>>Aerospace warning includes the monitoring of man-made objects in space, and the detection, validation, and warning of attack against North America whether by aircraft, missiles, or space vehicles, through mutual support arrangements with other commands.<<23:22
DocScrutinizerallegedly playing chess is also one of their duties23:23
DocScrutinizer;-)23:23
wpwrakkristianpaul: i would worry more about the headless chicken23:51
--- Mon Mar 5 201200:00

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