00:22.43 | *** join/#qi-hardware juan64bits (~juan64bit@181.68.200.127) |
01:07.13 | *** join/#qi-hardware panda|x201 (~hzhang@2001:c08:3700:ffff::e7d) |
01:41.59 | *** join/#qi-hardware wolfspraul (~wolfsprau@114.241.252.90) |
01:52.35 | *** join/#qi-hardware wej (~j@95.211.10.3) |
03:20.47 | *** join/#qi-hardware wpwrak (~werner@234-171-231-201.fibertel.com.ar) |
03:23.26 | *** join/#qi-hardware Openfree` (~Openfreer@116.228.88.131) |
04:20.14 | *** join/#qi-hardware antgreen (~user@bas3-toronto06-1176449386.dsl.bell.ca) |
04:44.40 | wolfspraul | wpwrak: hey good evening :-) |
04:44.59 | wolfspraul | I am thinking about doing some home-made PCB experiments like I saw at your place... |
04:45.07 | wolfspraul | need to pick a use case first though |
04:45.23 | wpwrak | whee ! :) |
04:45.30 | wolfspraul | in terms of packages, what do you think is the highest density/smallest pitch you would feel comfortable putting on a 1-layer PCB? |
04:45.46 | wolfspraul | obviously say if you take our slx45 fgg484 package, that won't work |
04:46.13 | wolfspraul | 0.5mm bga, 484 pins - probably the pitch is too small, and there are also far too many wires to route anywhere |
04:46.16 | wolfspraul | so that's obvious |
04:46.24 | wolfspraul | but what are the criteria you feel most comfortable with? |
04:46.33 | wolfspraul | qfn/qfp packages? so the wires only come out on the 4 sides? |
04:46.53 | wpwrak | BGA have the problem of pins/balls being inaccessible |
04:47.06 | wolfspraul | and also routing density I would think, no? |
04:47.17 | wolfspraul | you are working with 1-layer and 2-layer also now? |
04:47.47 | wpwrak | yes, QFN is the hardest i try my hands on. level of difficult depends a bit on the package shape. some have vertical walls and recessed pads. these are a bit difficult to solder. |
04:48.02 | wpwrak | 1 and 2 layer |
04:48.21 | wolfspraul | including drilling? |
04:48.37 | wpwrak | and yes, BGA may need multilayer. depends a bit on what it does |
04:48.43 | wpwrak | sure, with drilling |
04:48.49 | wolfspraul | ok, so qfn/qfp probably exchangable in terms of doability in your process |
04:48.55 | wolfspraul | what minimum pitch? |
04:49.15 | wolfspraul | have you tried 0.5mm spacing? |
04:49.28 | wolfspraul | or more like 2mm? :-) |
04:49.29 | wpwrak | but i can't plate the holes. i stick a little wire through them, but that means i can't have vias under small components |
04:50.59 | wpwrak | 0.45 mm QFN |
04:51.03 | wolfspraul | wow |
04:51.06 | wolfspraul | nice |
04:51.19 | wpwrak | (for the little UBB-like board with an atmel on it) |
04:51.45 | wolfspraul | ok there is a tqfp spartan-6, slx9 |
04:51.58 | wolfspraul | maybe my first goal is that and then see whether I can get parts of milkymist to run on it |
04:52.05 | wolfspraul | but the idea is (for me) the pcb process :-) |
04:52.17 | wpwrak | this one: http://downloads.qi-hardware.com/people/werner/uart/pix/uart-20110131.jpg |
04:52.27 | wpwrak | http://downloads.qi-hardware.com/people/werner/uart/pix/uart-inserted.jpg |
04:52.33 | wolfspraul | that slx9 has a 0.5mm spacing tqfp package |
04:53.21 | wpwrak | to get started, i'd recommend something with legs. tqfp, ssop, etc. |
04:54.07 | wpwrak | once have developed an intuition for the soldering, then you can tackle qfn |
04:56.42 | wolfspraul | you solder by hand or in a reflow oven? |
04:57.08 | wpwrak | for making the pcb, i cheat a little by letting my mill do all the drilling and cutting. but you can accomplish the same manually. it's just more fun to say "make drill" and off it goes, making those dozens of pesky little holes :) |
04:57.24 | wpwrak | by hand. oven didn't work too well. |
04:57.57 | wpwrak | and i tin the boards first, to make the soldering easier |
04:58.23 | wpwrak | tinning = apply flux, then "paint" tiny amounts of solder to just cover the copper |
04:59.11 | wpwrak | the tin also protects the copper from oxidation |
05:02.03 | wolfspraul | how do you paint? what solder? |
05:03.08 | wpwrak | "paint" = put a bit of solder on a the copper (preferably some large area), then move the iron such that a large surface/edge of the tip is parallel to the board |
05:03.34 | wpwrak | the resulting layer should be very thin and with even thickness |
05:03.54 | wpwrak | s/on a the copper/on the copper/ |
05:04.57 | wpwrak | solder doesn't matter that much. you can use lead-free or 63/37. also 60/40 isn't too bad. |
05:05.47 | wpwrak | what's very important is the flux |
05:07.05 | wpwrak | i use "water-soluble" flux, which has characteristics between RA and "no clean" |
05:08.26 | wpwrak | i find that "no clean" isn't very effective. and it evaporates quickly. and you still have to clean, or risk spurious connections in the 100 kOhm range |
05:09.26 | wolfspraul | ok! |
05:09.35 | wolfspraul | you flatten the solder with an iron |
05:09.41 | wpwrak | RA, on the other hand, does make the tin flow very well. it's also very sticky, so the components don't shift easily (you still have to hold them down when soldering, though). but the board pretty quickly becomes a sticky mess with RA. |
05:10.31 | wpwrak | with the soldering iron |
05:14.21 | wpwrak | for cleaning, you need a brush - a toothbrush is probably fine, but there are slightly fancier brushes with hog hair - and alcohol. if you want cleaning to be fun, look for an ultrasonic bath. they're not very expensive. |
05:15.54 | wpwrak | there is a lot of cleaning: after the tinning, after the soldering of the vias, after the soldering of components, and then after fixes. whenever the boards gets too messy, you want to clean off the flux |
05:31.18 | *** join/#qi-hardware jekhor (~jek@46.53.195.29) |
06:13.56 | *** join/#qi-hardware phirsch (~smuxi@xdsl-89-0-94-180.netcologne.de) |
06:26.11 | *** join/#qi-hardware xiangfu (~xiangfu@fidelio.qi-hardware.com) |
09:06.08 | *** join/#qi-hardware kristoffer (~kristoffe@c-c3dde555.010-30-6c6b7012.cust.bredbandsbolaget.se) |
10:12.24 | whitequark | +1000 on ultrasonic bath |
10:12.27 | whitequark | it's awesome |
10:13.29 | whitequark | wpwrak: on "painting" (isn't it called "tinning"?) I heard of an interesting method |
10:14.36 | whitequark | that boils down to using a spare pan, some water, a bit of citric acid and solder with a very low melting point |
10:14.59 | whitequark | an SnSbBi alloy I think |
10:15.48 | whitequark | you just boil it (pun intended) and place the board in the pan, where the alloy wets the copper (citric acid removes the oxide layer) |
10:16.39 | whitequark | then you pick a squeegee and remove excess tin |
10:17.35 | whitequark | it's very fast, low-temp and gives a nice flat surface |
10:17.59 | whitequark | of course, it's not Pb-free, but you know what? I won't drop a tear if all joints in the world would become Pb-containing. |
10:43.46 | *** join/#qi-hardware jurting (~jommpa90_@c213-100-125-198.cust.tele2.se) |
10:55.41 | *** join/#qi-hardware Ayla (~paul@114.46.123.78.rev.sfr.net) |
11:06.59 | *** join/#qi-hardware kilae (~chatzilla@catv-161-018.tbwil.ch) |
12:29.39 | *** join/#qi-hardware xiangfu (~xiangfu@123.114.251.98) |
12:47.09 | *** join/#qi-hardware xiangfu (~xiangfu@fidelio.qi-hardware.com) |
12:56.20 | *** join/#qi-hardware DocScrutinizer (~halley@openmoko/engineers/joerg) |
12:59.20 | *** join/#qi-hardware antgreen (~user@out-on-199.wireless.telus.com) |
13:13.09 | wpwrak | whitequark: (tinning) yes, that's the correct name for the process. the "painting" describes the method |
13:13.51 | wpwrak | (boiling lead plus acid bath) i'm not sure if wolfgang is up for that much chemistry ;-) |
13:20.37 | whitequark | erm, there's no boiling lead |
13:20.39 | whitequark | it's done in water |
13:20.46 | whitequark | with a diluted citric acid solution |
13:21.01 | whitequark | you could drink it (if there were no heavy metals in it, of course) |
13:21.12 | whitequark | I'd say it is somewhat safer than the soldering iron method. |
13:21.39 | whitequark | through yes, it requires a bit of preparation (squeegee, alloy, a spare pan) so maybe it's not for the first time |
13:26.00 | wpwrak | ah, the water gets in touch with the board. interesting. haven't heard of that one yet. the "standard" process uses some vile (cold) chemistry. |
13:39.29 | wolfspraul | I had some more questions, but what was it? :-) |
13:40.02 | wolfspraul | ah, one was whether you start with a plain FR4 material and then add the copper yourself? or you buy the FR4 already with copper and then etch? |
13:40.36 | wolfspraul | and the other one was whether you ever ran into the need, or found a solution how you can create stable (big) ground and power planes? |
13:43.00 | wpwrak | you get the board with the copper on it. then you etch away what you don't need |
13:43.51 | wpwrak | (big planes) well, you can use filled areas. with only two layers, whether your ground plane will be very contiguous depends a bit on the routing |
13:44.29 | wpwrak | and yes, i did run into the need with ben-wpan :) well, they're bit in relation to the size of the board, not in absolute terms |
13:45.49 | wolfspraul | filled areas? |
13:45.50 | wpwrak | to exercise the whole workflow, you may want to try something very simple. e.g., an MCU with a button and 1-2 LEDs. hook it up for programming to a Ben. |
13:46.15 | wolfspraul | could you make the bottom side a ground or power plane? |
13:46.34 | wolfspraul | you definitely cannot have a separate ground and power plane, and then still have a third one for the signal carrying wires |
13:46.58 | wpwrak | you can tell kicad that it should copper-fill an area (delimited by a non-self-intersecting polygon, optionally with holes). it will then leave room around traces but leave copper in the rest of the area |
13:47.07 | *** join/#qi-hardware GNUtoo-netbook (~gnutoo@host63-133-dynamic.20-79-r.retail.telecomitalia.it) |
13:47.28 | wpwrak | if you have an UBB at hand, you'll see that you can make a ground plane :) |
13:47.44 | wpwrak | or look at atben or atusb. the bottom is mostly ground |
13:48.04 | wolfspraul | you just used the entire bottom side as gnd plan there? |
13:48.31 | wpwrak | i tell kicad to use more or less all of the bottom for ground. then it figures out what to do |
13:48.39 | wolfspraul | you say "with copper on it" - on both sides? |
13:49.07 | wolfspraul | which fr4 thicknesses have you worked with? and which copper thickness? |
13:49.26 | wpwrak | e.g., it may not be able to connect all the things to ground, if there are other traces in the way. you usually need a few tries with routing changes until you get the ground where you want it |
13:50.03 | wpwrak | copper on both sides. i think the standard is "1 oz" (per square foot or such) |
13:50.21 | wpwrak | i used 1.6 mm, 0.8 mm, and 0.4 mm FR4 |
13:51.08 | wpwrak | 0.4 mm is fragile. only for things that need to be super-thin and have zero mechanical stress. |
13:51.37 | wpwrak | ubb, atben, and atusb are 0.8 mm |
13:51.58 | wpwrak | labsw is 1.6 mm (larger board, heavier components) |
13:52.28 | wpwrak | 1.6 mm is the "standard" thickness for DIY |
13:53.42 | wolfspraul | got it |
13:55.06 | wpwrak | 0.8 mm is friendlier for drilling. only half as far to go. you'll break fewer drill bits ;-) |
13:55.14 | wolfspraul | just out of curiosity. if you ever wanted to make a 3 or 4-layer pcb, could you just glue 2 thin ones together? |
13:55.36 | wpwrak | i would go to a pcb house :) |
13:56.17 | wpwrak | or figure out a way to reduce this to a regular 2 layer problem. maybe stack two boards |
13:56.39 | wpwrak | or use patch wires ;-) |
13:56.51 | wolfspraul | sure, I was just wondering whether you ever thought about stacking pcbs... |
13:57.30 | wolfspraul | I never walked through all the manufacturing steps of a high-end pcb, neither in reading nor an actual factory visit |
13:57.33 | wolfspraul | too bad |
13:57.36 | wpwrak | thought yes. the rejected :) |
13:58.36 | wolfspraul | yeah, of course. the idea is to find a solution with a simple pcb, not to compete with a pcb factory |
13:58.37 | wpwrak | i once looked up the industrial process for multilayer. a bit scary. needs high precision and a lot of steps. |
14:02.37 | wpwrak | have you seen this one yet ? http://www.almesberger.net/misc/led-toy/ledtoy-building-0pre9.pdf |
14:03.01 | wpwrak | this is a description of the all-manual process i used some six years ago |
14:03.59 | wpwrak | nowadays, i've automated a few things. and i added a few things (tinning, use of filled areas, etc.) |
14:16.26 | wpwrak | and of course i've changed the etching chemistry :) that ferric chloride is messy |
14:19.23 | wpwrak | HCl + H2O2 is much nicer to handle. the only drawback is that you have to be careful about chloride escaping (and eating any metals it can find) |
14:41.54 | whitequark | wolfspraul: I'm curious, for what do you need manual etching? curiousity? |
14:53.20 | *** join/#qi-hardware jurting (~jommpa90_@c213-100-125-198.cust.tele2.se) |
14:57.50 | *** join/#qi-hardware GNUtoo-desktop (~GNUtoo@host63-133-dynamic.20-79-r.retail.telecomitalia.it) |
15:00.40 | *** join/#qi-hardware zear (~zear@h196n1-g-kt-a31.ias.bredband.telia.com) |
15:15.29 | wolfspraul | to make pcbs? is that what you mean? |
15:16.52 | wolfspraul | you mean whether there are other ways to pattern the copper than via etching? |
15:39.16 | DocScrutinizer | yummy, H2O2 and HCl. only some Acetone missing |
15:40.06 | wpwrak | you can use that for wiping the toner off the PCB :) |
15:40.12 | wpwrak | (paint thinner will do too, though) |
15:40.26 | DocScrutinizer | :nod: |
15:40.41 | DocScrutinizer | APEX brings fun to your weekend |
15:41.01 | DocScrutinizer | devil's granny |
15:42.14 | DocScrutinizer | dang, now this channel moved all up to top of echelon's surveillance list |
15:43.03 | wpwrak | APEX ? |
15:43.33 | DocScrutinizer | ~wiki apex |
15:43.48 | DocScrutinizer | tsss |
15:44.01 | DocScrutinizer | Acetone PeroXide |
15:44.02 | wpwrak | nothing to see there |
15:44.31 | wpwrak | aah, good old TATP :) |
15:45.25 | DocScrutinizer | http://en.wikipedia.org/wiki/Acetone_peroxide |
15:48.04 | wpwrak | i should try to make some. there's a very annoying dog in the neighbourhood that i wouldn't mind blasting out of its miserable existence ... |
15:50.43 | wpwrak | hehe http://en.wikipedia.org/wiki/Acetone_peroxide#Accidental_byproduct |
15:51.35 | wpwrak | if you mix chemical waste of potent ingredients, you deserve whatever you get ;-) |
15:56.47 | DocScrutinizer | aaah >>adding a soluble copper(II) compound.<< for a measure to avoid TATP by-products in industry. Your etching might be safe ;-) |
15:59.52 | DocScrutinizer | fun bit: MDMA synthesis has potential to create TATP as byproduct ;-P |
16:01.24 | DocScrutinizer | the German slang "das zeug knallt" gets a very interesting new meaning |
16:04.17 | DocScrutinizer | ooh >>nickname "Mother of Satan"<< - not devil's grandma like I thought |
16:04.34 | GNUtoo-desktop | hi kristianpaul |
16:06.58 | DocScrutinizer | wpwrak: anyway, german wikipedia > english, as usual. See http://de.wikipedia.org/wiki/Apex |
16:07.27 | whitequark | DocScrutinizer: hm, interesting. I always notice that english wiki >> russian |
16:07.31 | DocScrutinizer | >>Apex oder APEX steht als Abkürzung für: |
16:07.32 | DocScrutinizer | Acetonperoxid, Explosivstoff<< |
16:07.51 | GNUtoo-desktop | DocScrutinizer, german wikipedia seem to have more on free software for GSM protocols such as openBSC for instance |
16:07.52 | whitequark | maybe I really should finally learn Deutsch |
16:09.24 | GNUtoo-desktop | what's the correct channel for getting help on things like: "does someone know how to tell if a PIC is dead or not, I tried that: https://sites.google.com/site/thehighspark/arduino-pic18f and it always returns FF FF etc....after reading" |
16:09.58 | whitequark | what |
16:10.01 | whitequark | pic18f isn't avr |
16:10.28 | GNUtoo-desktop | I've both |
16:10.30 | whitequark | ah, programmer. I should have read the link |
16:10.44 | GNUtoo-desktop | I want to program pic18f2550 from arduino |
16:10.44 | whitequark | well, not working properly is the normal state of pics |
16:10.48 | whitequark | enjoy your misery |
16:11.02 | GNUtoo-desktop | yes that's what I was told, I already tried to make it works during ages |
16:11.08 | GNUtoo-desktop | and I abandoned |
16:11.21 | GNUtoo-desktop | but here I found this new arduino-based programmer and I wanted to give a try |
16:11.26 | whitequark | try electronics.stackexchange.com |
16:11.43 | whitequark | (if you've already tried replacing chips |
16:11.44 | whitequark | ) |
16:11.46 | GNUtoo-desktop | or maybe I should try ##electronics then |
16:11.54 | GNUtoo-desktop | no, I've only one chip |
16:12.05 | whitequark | then it's probably just dead, or locked |
16:12.17 | GNUtoo-desktop | ok |
16:12.31 | GNUtoo-desktop | then I killed it before while trying to program it like one year ago |
16:12.39 | whitequark | wpwrak can tell you a LOT about spurious locking :) |
16:12.43 | DocScrutinizer | haha >> Bei mangelnder Kühlung entsteht bei der Reaktion mit Salzsäure unter Kochen das Tränengas Chloraceton.<< |
16:12.51 | whitequark | besides which, pics aren't even nearly as resilient as avrs |
16:13.28 | whitequark | GNUtoo-desktop: you connected 12V to MCLR, didn't you? |
16:13.52 | whitequark | if I recall it correctly, PICs require HV-programming |
16:14.10 | GNUtoo-desktop | whitequark, no, but there is low voltage programming nowadays |
16:14.26 | whitequark | ok |
16:14.33 | GNUtoo-desktop | but maybe I activated the 12V once and cannot go back to 5V? |
16:14.37 | whitequark | then no idea, I only us.. fucked with HV programming |
16:14.53 | whitequark | it was a huge PITA |
16:15.07 | whitequark | some bastard said that PICs are better for embedded developing, somewhere back in 2006 |
16:15.22 | whitequark | that was before arduinos and popularity of them across laymen |
16:15.28 | whitequark | grr |
16:15.42 | GNUtoo-desktop | ah the eev blog |
16:15.49 | whitequark | eev blog? |
16:15.52 | GNUtoo-desktop | yes he even hates gcc |
16:16.12 | GNUtoo-desktop | http://www.youtube.com/watch?v=DBftApUQ8QI |
16:16.29 | whitequark | in 2006 I didn't know enough english to read blogs :) |
16:16.36 | GNUtoo-desktop | ok |
16:16.42 | whitequark | so it was a russian bastard |
16:16.48 | GNUtoo-desktop | ok |
16:16.57 | GNUtoo-desktop | personally I buoght the pic for the pinguino |
16:17.04 | GNUtoo-desktop | because it was usb,faster than arduino etc... |
16:17.10 | GNUtoo-desktop | *native usb |
16:17.42 | whitequark | if you want usb and don't want to hunt for atusb's, listen to my advice: buy an STM32 series chip |
16:17.50 | GNUtoo-desktop | ok |
16:17.55 | whitequark | it's a Cortex-M3 ARM and it has an experience which is really close to AVRs |
16:17.57 | GNUtoo-desktop | that's even supported by nuttx |
16:18.55 | whitequark | it has very similar peripherals (but more of them), significantly more orthogonal registers (due to 32-bit addresses), some more interesting stuff like bitbanding areas |
16:19.04 | whitequark | it is 3V3-powered but 5V-tolerant |
16:19.30 | whitequark | but I doubt that 3V3 is a problem for anyone these days. even some arduinos are 3V-powered |
16:19.47 | whitequark | also, li-ion-compatible: works at least from 2.7V to 3.6V |
16:19.56 | whitequark | best of all |
16:20.25 | whitequark | they are significantly faster than AVRs (hardware multiplier for example, or 1-cycle 32-bit memory fetch), cheaper, and eat significantly less current |
16:21.07 | whitequark | an STM32 on low-freq oscillator like 32.767K can perform something useful and consume something like 0.5mA |
16:21.16 | whitequark | ideal for battery- or even solar-powered designs |
16:21.21 | GNUtoo-desktop | wow |
16:21.26 | whitequark | exactly |
16:21.34 | GNUtoo-desktop | about battery, what about ti launchpad? |
16:21.45 | whitequark | I seen a video (on a local blog) where an STM32V (even a more low-power version) is powered from an apple |
16:21.48 | whitequark | and drives a LCD |
16:22.03 | whitequark | from a damned _apple_. like, a fruit. with zinc and copper electrodes in |
16:22.05 | GNUtoo-desktop | launchpad have integrated gdb support..... |
16:22.07 | whitequark | for hours. |
16:22.15 | GNUtoo-desktop | ok |
16:22.17 | GNUtoo-desktop | wow |
16:22.19 | whitequark | STM32 has gdb in-circuit debugger too |
16:22.23 | GNUtoo-desktop | wow |
16:22.25 | whitequark | I added support for it |
16:22.29 | GNUtoo-desktop | nice |
16:22.33 | whitequark | github.com/texane/stlink |
16:22.36 | GNUtoo-desktop | ok |
16:23.51 | GNUtoo-desktop | stm32 is not as low cost as ti launchpad? |
16:23.55 | whitequark | in a very modest stm32 you get 8K ram, 128K rom, fast ARMv7 fully 32-bit core, 3x of peripherals compared to a typical atmega, which can also often be remapped to different ports |
16:23.59 | whitequark | hmm |
16:24.24 | whitequark | the dev-board with a chip I described above and an in-circuit debugger (which you can use for your own circuits too) costs ~$10 |
16:24.36 | whitequark | it's also breadboard-friendly. just plug in and go |
16:24.43 | GNUtoo-desktop | ok so it's low enough |
16:24.46 | whitequark | STM32VLDISCOVERY |
16:24.59 | whitequark | launchpad is msp430 |
16:25.09 | GNUtoo-desktop | for instance for doing remote temperature sensors I wonder what's best |
16:25.18 | whitequark | msp430 is a very old and quite crappy architecture. I think it dates back to 6502 or something like that |
16:25.21 | GNUtoo-desktop | maybe the bluecore even with the modified firmware |
16:25.32 | GNUtoo-desktop | ok |
16:25.33 | whitequark | GNUtoo-desktop: use STM32V, a low-power low-voltage version |
16:25.39 | GNUtoo-desktop | so not worth for msp430 |
16:25.45 | whitequark | there's even one with integrated radio transceiver |
16:25.46 | whitequark | 2.4g |
16:25.51 | whitequark | just add two caps and go |
16:25.51 | GNUtoo-desktop | wow |
16:25.55 | whitequark | yes |
16:26.04 | whitequark | let me find it |
16:26.06 | GNUtoo-desktop | ok |
16:26.15 | GNUtoo-desktop | is there some good kit at sparkfun? |
16:26.31 | GNUtoo-desktop | the problem is also that I already have an arduino |
16:26.46 | GNUtoo-desktop | but I don't like the arduino language, so I program it in avr-libc of course |
16:27.13 | whitequark | uh, sparkfun. I don't know much of it 'cause they only ship to Russia with fedex and it's insanely expensive. apart from that, they allow no more than $70 of declared value. |
16:27.29 | whitequark | given that shipping is ~$70 too, it's barely effective. |
16:29.51 | whitequark | stm32's also have "device" and "otg" mode USB, which means you can plug in peripherals |
16:30.18 | GNUtoo-desktop | ok |
16:30.35 | GNUtoo-desktop | yes I think it also does host |
16:30.40 | whitequark | check STM32W |
16:30.43 | whitequark | series |
16:30.48 | whitequark | consumes 0.25mA/MHz |
16:30.50 | GNUtoo-desktop | if I remember well what I saw on nuttx mailing list |
16:30.54 | whitequark | and has 1.25DMIPS/MHz :D |
16:31.31 | whitequark | http://www.st.com/internet/mcu/product/245419.jsp |
16:31.38 | GNUtoo-desktop | http://www.sparkfun.com/search/results?term=stm32&what=products |
16:31.39 | whitequark | I guess they have a zigbee stack handy for this chip |
16:31.42 | GNUtoo-desktop | that's all they have |
16:32.07 | whitequark | yeah, 802.15.4 |
16:32.17 | GNUtoo-desktop | yes I saw, |
16:32.27 | GNUtoo-desktop | but I wonder if there is bluetooth |
16:32.28 | whitequark | the chip is pretty new and I think that sparkfun doesn't have boards (yet?) |
16:32.37 | GNUtoo-desktop | ok |
16:32.43 | whitequark | er, digikey has one for $350 |
16:32.49 | GNUtoo-desktop | ouch |
16:32.50 | whitequark | this is a bit too much for my taste |
16:32.54 | GNUtoo-desktop | me too |
16:33.05 | GNUtoo-desktop | less than 50E is the maximum |
16:33.16 | whitequark | hm |
16:33.29 | GNUtoo-desktop | *about |
16:33.44 | DocScrutinizer | sounds quite nice anyway |
16:33.53 | whitequark | if I'd have this task, I would copy the relevant part of layout, send it to a pcb fab and have it manufactured for ~$20 |
16:34.09 | whitequark | then just buy the chip, it's ~$3 |
16:34.12 | whitequark | and solder it |
16:34.18 | DocScrutinizer | YOU |
16:34.25 | whitequark | huh? |
16:34.36 | DocScrutinizer | not joe average though :-) |
16:34.47 | whitequark | I guess GNUtoo isn't an joe average too :) |
16:34.56 | DocScrutinizer | who knows :-D |
16:34.59 | GNUtoo-desktop | depends, I had a hard time with the pic |
16:35.19 | GNUtoo-desktop | and I don't want to solder too small stuff |
16:35.26 | DocScrutinizer | see? |
16:35.34 | GNUtoo-desktop | but a big enough thing is ok |
16:35.35 | whitequark | huh, well it kind of oversatisfies the definition of "small" |
16:35.39 | whitequark | 4x4mm |
16:35.40 | whitequark | or so |
16:35.48 | whitequark | and 48 pads |
16:35.52 | DocScrutinizer | yay |
16:35.53 | whitequark | it's pretty easy in fact |
16:36.10 | whitequark | you just slap some paste, then slap the chip over it and heat |
16:36.17 | DocScrutinizer | yup |
16:36.24 | GNUtoo-desktop | for instance the 28pin PDIP SOIC pic is ok |
16:36.27 | DocScrutinizer | if you got a nice hot-air reflow |
16:36.59 | whitequark | GNUtoo-desktop: ah, another fact: there are no STM32's in DIP |
16:37.07 | whitequark | actually, there's just one ARM ever in DIP and it sucks |
16:37.42 | whitequark | but I'd say that TQFPs are almost as easy when you overcome initial fear of "tiny parts" |
16:37.43 | GNUtoo-desktop | but for instance the one used in the tricorder is not ok |
16:37.48 | whitequark | tricorder? |
16:37.48 | DocScrutinizer | DIP is kinda out |
16:38.14 | whitequark | DocScrutinizer: yeah, I'm waiting until it finally dies and everyone will understand that TQFP-is-not-hard-at-all |
16:38.14 | DocScrutinizer | though I still love my AT89C2051 design |
16:38.16 | GNUtoo-desktop | http://www.tricorderproject.org/tricorder-mark2.html |
16:38.23 | whitequark | not going to happen in 50 yrs through |
16:38.27 | whitequark | DocScrutinizer: ewww, 8051 |
16:38.56 | whitequark | GNUtoo-desktop: that's a TQFP or LQFP, I couldn't quite recognize pitch from the photo |
16:38.58 | DocScrutinizer | well, it's just another assembler |
16:39.07 | GNUtoo-desktop | http://www.tricorderproject.org/tricorder_pictures/mark2/tricorder_mk2_motherboard2.jpg |
16:39.09 | whitequark | DocScrutinizer: well, even 8051 is better than PIC |
16:39.25 | whitequark | looks like tqfp-144 |
16:39.26 | DocScrutinizer | nice DIP, with IO that drive LED etc directly (up to 20mA iirc) |
16:39.46 | viric | what is tqfp? |
16:39.55 | whitequark | Thin Quad Flat Package |
16:39.58 | whitequark | a form of chip cases |
16:40.00 | whitequark | *form-factor |
16:40.01 | viric | ah |
16:40.03 | whitequark | http://en.wikipedia.org/wiki/TQFP |
16:41.26 | whitequark | a Ruby guy (@headius, lead of jruby project actually) once asked something about PICs and I accidentally slipped into an uncontrollable rage |
16:41.37 | whitequark | so there's one PIC user less in this world |
16:41.54 | whitequark | each day spent fighting with PICs will be remembered by our children |
16:41.57 | viric | you killed him? |
16:42.01 | whitequark | no |
16:42.19 | whitequark | but he was reading it for half a hour or so |
16:42.43 | whitequark | and kind of decided not to try anything that can cause such a psychological damage |
16:43.06 | GNUtoo-desktop | lol qualcomm should also be blacklisted along with pics |
16:43.11 | whitequark | seriously, MEMORY BANKS |
16:43.26 | whitequark | it's fucking 2012 (and it was fucking 2006 when I first encountered them) |
16:43.34 | whitequark | where spare SRAM is poured across registers |
16:43.44 | whitequark | and PORTx registers aren't safe to |= |
16:43.53 | GNUtoo-desktop | ouch |
16:43.56 | whitequark | YES. |
16:44.04 | whitequark | not to mention it works on 20mhz max |
16:44.10 | whitequark | and it takes FOUR cycles to execute a command |
16:44.36 | whitequark | and it has 12-bit-wide instruction data bus |
16:44.43 | whitequark | so the firmware is like 0xfff 0xfff 0xfff |
16:45.13 | GNUtoo-desktop | lol ouch |
16:45.21 | whitequark | I guess its developer was extremely sex-deprived or something like that, and so this beast of sublimation was born |
16:45.26 | whitequark | no idea how to explain this other ways |
16:45.39 | GNUtoo-desktop | patents? |
16:45.44 | GNUtoo-desktop | like to avoid patents? |
16:45.48 | whitequark | memory banks mean that you have two bits that control which of four banks is selected |
16:46.00 | whitequark | and some registers and SRAM cells are in all banks |
16:46.08 | whitequark | some are in only one |
16:46.19 | whitequark | and you gotta use two instructions to switch to a particular bank |
16:46.22 | whitequark | TWO. |
16:46.48 | whitequark | so you need to reorder your code carefully to avoid unnecessary switches which eat your cpu time and program storage (it's also quite small) |
16:47.17 | whitequark | hence, no gcc and no proper C compiler at all (except for their proprietary crap which still produces bad code because you just cannot do anything good with this braindead architecture) |
16:47.18 | GNUtoo-desktop | ok |
16:47.27 | whitequark | sdcc is... tolerable |
16:47.41 | GNUtoo-desktop | ahh that's why there is no GCC and only SDCC |
16:47.41 | whitequark | if you can call putting a hand into a hive "tolerable", that is |
16:48.06 | whitequark | it's "tolerable" like "maybe you won't be dead in 8 hours but will be in 12" |
16:48.15 | GNUtoo-desktop | lol |
16:48.19 | whitequark | seriously, I'd rather put my hand into a hive. |
16:48.45 | whitequark | I once read a very nice tale about a programmer (in Russian, unfortunately) |
16:48.48 | GNUtoo-desktop | like theses people that do extremly dangerous stuff on TV to become famous? |
16:49.03 | whitequark | it was about a programmer who lost his mind trying to shrink his program by one byte |
16:49.10 | GNUtoo-desktop | ouch |
16:49.26 | GNUtoo-desktop | personally I started loosing my mind with the htc dream |
16:49.34 | whitequark | it wasn't (supposedly) real, just a nice piece of fictional writing |
16:49.41 | GNUtoo-desktop | ok |
16:49.45 | whitequark | or maybe it was, I dunno, but the author writes really good |
16:49.49 | whitequark | so, I recently re-encountered it |
16:49.51 | GNUtoo-desktop | ok |
16:49.59 | whitequark | and from numerous references it was obvious he was trying to code for PIC |
16:50.04 | GNUtoo-desktop | the effect of losing my mind is nervousness |
16:50.05 | whitequark | YES IT IS LIKE SO' |
16:50.23 | GNUtoo-desktop | ok lol |
16:50.33 | whitequark | I think I had exactly the same PIC he had, at least it was definitely the same core and probably series |
16:50.46 | GNUtoo-desktop | ok |
16:50.56 | whitequark | argh now I have an urge to kill someone. |
16:51.14 | GNUtoo-desktop | can you try on steeve balmer? |
16:51.37 | whitequark | steve ballmer is a saint compared to developer who invented pic |
16:51.43 | GNUtoo-desktop | ok |
16:51.46 | whitequark | at least he doesn't force people to be tortured |
16:51.49 | whitequark | well he does |
16:51.54 | GNUtoo-desktop | lol |
16:52.18 | whitequark | but pics are worse than that. it's like tearing your own nails with red-hot tweezers and then eating them |
16:52.21 | whitequark | for a week |
16:52.21 | whitequark | one nail a day |
16:52.37 | GNUtoo-desktop | ok |
16:52.58 | whitequark | I hope I convinced you not to use PICs :) |
16:53.04 | GNUtoo-desktop | so I think I won't try to use the pic I have nor buy a newer one even if it's just a jump to the electronic shop |
16:53.28 | GNUtoo-desktop | yes I've already enough nervousness issues..... |
16:53.56 | GNUtoo-desktop | just undid his pic circuit on his breadboard |
16:59.20 | whitequark | if you want to feel what I think, just imagine that everything above was said with voice and style of Freeman's Mind |
16:59.23 | whitequark | http://www.youtube.com/watch?feature=player_detailpage&v=ayeCckhGJ-c#t=443s |
17:01.35 | GNUtoo-desktop | maybe you should go kill http://www.youtube.com/watch?v=DBftApUQ8QI who promotes PICs |
17:02.27 | whitequark | yeah, I'll just get my submachine gun |
17:06.23 | *** join/#qi-hardware GNUtoo-netbook (~gnutoo@host63-133-dynamic.20-79-r.retail.telecomitalia.it) |
17:31.32 | *** join/#qi-hardware jurting (~jommpa90_@c213-100-125-198.cust.tele2.se) |
17:35.42 | kristianpaul | hola GNUtoo-desktop |
17:35.55 | GNUtoo-netbook | hi |
17:36.34 | GNUtoo-netbook | I just wanted to retry with the pic but whitequark convinced me not to (I found a pic programmer made from an arduino...) |
17:39.01 | GNUtoo-netbook | beside I do not like the integration of most programmer done with arduino, because they are not compatible with the standard flashing tools like the picprog,avrdude or flashrom |
17:39.40 | GNUtoo-netbook | that made me make my own one for flashing BIOS/Coreboot in avr-libc(and it even appeared on hackaday) |
17:40.38 | GNUtoo-netbook | (http://hackaday.com/2012/05/04/arduino-uno-bios-flasher/) |
17:40.50 | GNUtoo-netbook | kristianpaul, are you still doing the GPS thing? |
17:41.16 | kristianpaul | yes |
17:41.17 | GNUtoo-netbook | kristianpaul, 2) pinguino IRC channel seem to have disapeared.... |
17:41.19 | GNUtoo-netbook | ok |
17:41.35 | GNUtoo-netbook | kristianpaul, I'm also into basebands now |
17:41.40 | kristianpaul | perhaps, but ML is more active |
17:41.42 | GNUtoo-netbook | ok |
17:42.38 | kristianpaul | are coding a baseband? for wich app? |
17:42.41 | GNUtoo-netbook | (nuttx port to osmocombb phones) |
17:43.03 | kristianpaul | ah yes, |
17:43.16 | kristianpaul | that will cool |
17:43.31 | GNUtoo-netbook | we got a good part upstream |
17:44.41 | kristianpaul | do you have a gui now? |
17:45.08 | GNUtoo-netbook | not yet |
17:45.23 | GNUtoo-netbook | but we have CLI and poweroff and SPI |
17:46.05 | GNUtoo-netbook | http://bb.osmocom.org/trac/wiki/nuttx-bb/drivers |
17:48.30 | GNUtoo-netbook | we used initial code tough |
17:51.20 | GNUtoo-netbook | what;s the status of your GPS project? |
17:52.54 | kristianpaul | Artyom from gnss-sdr.ru had get to work tracking using a maxim based board |
17:53.02 | kristianpaul | right now there are two plans, |
17:53.23 | kristianpaul | 1) Adapt board from gnss-sdr.ru to milkymist R3 |
17:53.59 | kristianpaul | 2). Port osgps (gpsrcv.c) to a baremetal milkymist app orr rtems, whatever got working first |
17:54.24 | GNUtoo-netbook | ok |
17:54.33 | kristianpaul | having those thigns done your got a full hardware and software gps receiver |
17:54.37 | GNUtoo-netbook | wow |
17:54.50 | kristianpaul | do you want to help us to port osgps? |
17:55.33 | kristianpaul | is mostl plubling and some fixes to make it work with out libc :) (if baremetal way is choosen) |
17:55.39 | kristianpaul | most/mostly |
17:55.49 | GNUtoo-netbook | hmmm |
17:56.12 | GNUtoo-netbook | I am ultra busy..... |
17:56.18 | kristianpaul | right now i'm learning a bit kicad to make board adaptation |
17:56.33 | kristianpaul | sure i'm kinda aware, but i cant hold my self and ask you help |
17:56.39 | kristianpaul | since you mentioned baseband workds |
17:56.43 | GNUtoo-netbook | ok |
17:57.01 | GNUtoo-netbook | I'll look after eating |
17:57.21 | kristianpaul | s/workds/words |
17:57.21 | qi-bot | kristianpaul meant: "since you mentioned baseband words" |
17:59.00 | kristianpaul | about osgps, we dont need warm-start suport right now, so basically this is the major modification need to be done |
17:59.19 | *** join/#qi-hardware Ayla_ (~paul@75.165.26.93.rev.sfr.net) |
18:00.36 | GNUtoo-netbook | ok |
18:01.09 | whitequark | oh, I have a question about gps |
18:01.58 | whitequark | I have a smartphone with SiRF GPS chip, and sometimes it takes quite a long time to lock onto the signal in spite of clear sky view |
18:02.22 | whitequark | I read that GPS baseband needs to download almanac and ephemerides, which are transmitted with 300 baud or something like that |
18:02.32 | whitequark | am I right that it's what it does? |
18:02.43 | kristianpaul | yes |
18:02.57 | whitequark | where is it generally stored? |
18:03.00 | kristianpaul | that could take up to 30minutes |
18:03.04 | whitequark | does the chip have a ROM/flash? |
18:03.07 | kristianpaul | in your baseband chip :) |
18:03.10 | kristianpaul | of course |
18:03.24 | whitequark | well, GSM baseband for example doesn't have its own memory here |
18:03.43 | whitequark | it stores calibration data on the main flash via a RPC interface to the kernel |
18:03.46 | whitequark | and RIL |
18:03.53 | kristianpaul | if it support warmstart is should have embeded some flash yes |
18:04.02 | whitequark | ok |
18:04.15 | kristianpaul | i havent dig about all current avaliable gps receivers |
18:04.21 | kristianpaul | but basically thats a SoC |
18:04.36 | kristianpaul | embeded with some IP cores perhaps fir fancy FFT, dont know |
18:05.13 | kristianpaul | but if it speaks SiRF or NMEA is like a GSM Modem speaking AT |
18:06.24 | GNUtoo-netbook | let me read |
18:06.31 | kristianpaul | whitequark: btw agps could provide some speedup on this |
18:06.35 | kristianpaul | i havent read in depth |
18:07.03 | kristianpaul | guess is simlar to RRLP |
18:08.17 | whitequark | kristianpaul: I tried sending AGPS update command |
18:08.20 | whitequark | no effect actually |
18:08.30 | GNUtoo-netbook | SIRF is supported by gpsd |
18:08.32 | whitequark | ah, the gps baseband is embedded to gsm baseband actually |
18:08.37 | whitequark | that's why, I understand it now |
18:08.46 | kristianpaul | it could yes |
18:09.10 | GNUtoo-netbook | what;s the soc? qualcomm? |
18:09.12 | kristianpaul | ah yours is a gsm + gps combo? |
18:09.47 | whitequark | kristianpaul: yeah, according to lvl3 servicemanual |
18:10.11 | kristianpaul | interesting you can rum agps updates, |
18:10.23 | kristianpaul | thre is a debug command about correlation or counters? |
18:10.36 | whitequark | kristianpaul: no idea, I just clicked a button in GPSTest app :) |
18:10.41 | whitequark | I can give you shell |
18:10.45 | kristianpaul | ahh |
18:10.46 | kristianpaul | k |
18:11.01 | whitequark | GNUtoo-netbook: IIRC yes |
18:11.33 | GNUtoo-desktop | ok what's the phone? |
18:11.39 | whitequark | SGS2 |
18:11.42 | GNUtoo-desktop | ok |
18:11.50 | GNUtoo-desktop | ah so samsung protocol |
18:12.18 | GNUtoo-desktop | whitequark, 2 news on SGS2: replicant developer paulk will have an SGS2 |
18:12.29 | GNUtoo-desktop | whitequark, shared memory is probably over dual port ram |
18:12.59 | whitequark | GNUtoo-desktop: well it was pretty obvious about dualport ram |
18:13.03 | whitequark | what other options do you have? |
18:13.07 | GNUtoo-desktop | HSI |
18:13.18 | GNUtoo-desktop | that's what there is on nexus prime |
18:13.22 | whitequark | High-Speed Interconnect? |
18:13.31 | GNUtoo-desktop | high speed serial |
18:13.32 | whitequark | (just guessing) |
18:13.34 | whitequark | ah |
18:13.39 | whitequark | so it's not quite RAM, right? |
18:13.45 | GNUtoo-desktop | indeed |
18:13.52 | GNUtoo-desktop | that's omap's high speed serial port |
18:14.00 | whitequark | ewww, ti socs |
18:14.18 | whitequark | through omaps may be better than the stuff I've seen. |
18:14.19 | GNUtoo-desktop | even some versions of SGS2 have TI soc |
18:15.41 | whitequark | I'm not sure that exynos is significantly better anyway. Docs are definitely lacking |
18:17.06 | GNUtoo-desktop | we have leaked docs for s5pc110 at least |
18:17.22 | GNUtoo-desktop | +samsung is a big linux kenrel contributor |
18:18.10 | GNUtoo-desktop | but for instance the big problem in the samsung soc version is the CODEC |
18:18.46 | whitequark | nah, I wasn't talking about their openness or such. I was talking about general quality of resources, that is, I'm pretty sure that their internal ones are shitty too |
18:18.56 | whitequark | it seems to be a general problem with this insane race |
18:19.17 | GNUtoo-desktop | but in another hand the tizen project has a clean implementation of SGS2 CODEC |
18:19.28 | whitequark | can't it be merged? |
18:19.35 | GNUtoo-desktop | it can |
18:19.42 | whitequark | nice |
18:19.58 | whitequark | any news on Lima? |
18:20.03 | GNUtoo-desktop | just that Alex[sp3dev] just found out the other day |
18:20.17 | GNUtoo-desktop | (he's doing the GNU/Linux port on SGS2) |
18:20.21 | whitequark | that's pretty exciting. the phone is barely useful without proper acceleration due to battery drainage |
18:20.29 | GNUtoo-desktop | ah? |
18:20.35 | whitequark | [Lima is pretty exciting] |
18:20.42 | GNUtoo-desktop | ah was for battery drainage |
18:20.58 | GNUtoo-desktop | yes |
18:21.07 | *** join/#qi-hardware Aylax (~Aylax@90.84.144.167) |
18:21.09 | whitequark | well, Android <4 is long-term irrelevant |
18:21.13 | GNUtoo-desktop | there is also qualcomm 3d comming, but I'm waiting for TI 3d |
18:21.20 | whitequark | I'm already on ICS and it's the first sensible version of android |
18:21.23 | whitequark | 2.x feels like windows 98 |
18:21.35 | GNUtoo-desktop | ah ok |
18:21.38 | GNUtoo-desktop | I never tried ICS |
18:22.01 | whitequark | try it; you'll remember 2.x as a horrible nightmare |
18:22.07 | GNUtoo-desktop | kristianpaul, what were the tasks with osgps and what's the required hardware? |
18:22.18 | GNUtoo-desktop | whitequark, I'm in a worsenightmare: gta02 |
18:22.29 | GNUtoo-desktop | I tried during hours to make GPRS work the other day |
18:22.30 | whitequark | oh |
18:22.46 | GNUtoo-desktop | gta04 doesn't work for me since I have a3 version |
18:22.49 | kristianpaul | GNUtoo-desktop: gp2021 is implemented on hardware as HDL |
18:23.04 | whitequark | well, I can understand mucking with gta02 due to freedom issues and such, but as a general phone it's unusable imo |
18:23.04 | GNUtoo-desktop | which requires an alsa forwarder to be done |
18:23.06 | kristianpaul | GNUtoo-desktop: osgps bases its work from it, basically it is memory maped |
18:23.19 | GNUtoo-desktop | whitequark, part of it is usable part of it is not |
18:23.25 | kristianpaul | GNUtoo-desktop: dont bother fore required hardware for now, |
18:23.29 | GNUtoo-desktop | for instance GPRS is unusable |
18:23.37 | kristianpaul | you just need qemu for milkymist/lm32 |
18:23.42 | GNUtoo-desktop | ok |
18:23.47 | GNUtoo-desktop | bbs |
18:23.54 | kristianpaul | milkymist.org/wiki have the info |
18:24.54 | kristianpaul | despite having a not ending loop reading gp2021 register, osgps process cold start and display data on screen about position |
18:24.59 | kristianpaul | dopler |
18:25.06 | kristianpaul | and satellites on sky |
18:30.13 | Aylax | whitequark: it's that good? |
18:34.05 | *** join/#qi-hardware phirsch (~smuxi@xdsl-89-0-146-114.netcologne.de) |
18:35.21 | *** join/#qi-hardware mth (mxqegljt@dhcp-089-098-069-120.chello.nl) |
18:36.23 | *** join/#qi-hardware phirsch (~smuxi@xdsl-89-0-146-114.netcologne.de) |
18:39.10 | GNUtoo-desktop | ok |
18:47.36 | GNUtoo-desktop | hi DocScrutinizer |
18:48.09 | GNUtoo-desktop | the openmoko debug board has an FTDI chip |
18:48.42 | GNUtoo-desktop | it's supported too by flashrom for flashing SPI chips |
18:48.52 | GNUtoo-desktop | but no informations on the wiring.... |
18:52.52 | wpwrak | GNUtoo-desktop: for in-circuit programming, then Ben is remarkably useful. i use it for chips from the 8051, AVR, and PIC families |
18:53.37 | *** join/#qi-hardware jekhor (~jek@46.53.195.29) |
18:54.19 | kristianpaul | wpwrak: PIC? |
18:54.30 | kristianpaul | and you dont toll us !! |
18:58.48 | DocScrutinizer | HSI is *not* shared RAM, and it's also not a TI thing |
18:59.15 | GNUtoo-desktop | DocScrutinizer, indeed |
18:59.25 | GNUtoo-desktop | ah it's not a TI thing? |
19:00.01 | GNUtoo-desktop | usually silicon vendor pick some "IP" blocks, assemble them and do a chip |
19:01.10 | GNUtoo-desktop | so maybe that's because of that that it's not a TI thing |
19:01.41 | DocScrutinizer | STE is using HSI as well |
19:02.03 | DocScrutinizer | Thot CAIF is usually over HSI |
19:02.07 | DocScrutinizer | Thor* |
19:02.27 | DocScrutinizer | it's a interface just like I2C or SPI |
19:02.38 | DocScrutinizer | just incredibly hard to find specs for it |
19:02.53 | GNUtoo-desktop | ok |
19:03.38 | DocScrutinizer | MISI-HSI ??? |
19:03.48 | DocScrutinizer | something like that |
19:04.11 | DocScrutinizer | MIPI-HSI |
19:04.32 | DocScrutinizer | http://www.mipi.org/specifications/high-speed-synchronous-serial-interface-hsi |
19:04.32 | GNUtoo-desktop | ah ok |
19:04.34 | GNUtoo-desktop | thanks a lot |
19:04.40 | GNUtoo-desktop | yes I ddg it and found that |
19:04.53 | GNUtoo-desktop | Specifications are available to MIPI members only. For more information on joining MIPI, please go to Join MIPI. |
19:04.55 | GNUtoo-desktop | hmmm |
19:05.19 | wpwrak | kristianpaul: aye, PIC :) the architecture is evil but you get to make all sorts of PIG jokes :) |
19:06.18 | DocScrutinizer | yeah, I of course have seen the datasheet |
19:06.48 | *** join/#qi-hardware emeb (~ericb@ip72-223-81-94.ph.ph.cox.net) |
19:13.49 | DocScrutinizer | http://www.mail-archive.com/linux-omap@vger.kernel.org/msg18506.html |
19:14.17 | GNUtoo-desktop | yes I'm aware of that patch |
19:15.18 | DocScrutinizer | nfc what's this: http://www.scribd.com/doc/44735645/72/MIPI-HSI-INTERFACE-CONTROLLER |
19:15.29 | DocScrutinizer | I have no facebook account |
19:16.13 | DocScrutinizer | anyway HSI is a 4wire bidir interface, with up to (iirc) 200Mb/s |
19:16.47 | DocScrutinizer | current version is 1.2, though a lot of devices only support 1.0.1 |
19:16.57 | GNUtoo-desktop | ok |
19:18.05 | DocScrutinizer | it's genuinely supporting logical channels |
19:18.10 | DocScrutinizer | iirc up to 8 |
19:19.19 | DocScrutinizer | some very tricky stuff with long and short addresses in data package headers |
19:19.38 | GNUtoo-desktop | ok |
19:20.21 | DocScrutinizer | you definitely don't want to touch internals if you don't have to and there'S any driver with API |
19:20.33 | GNUtoo-desktop | ok |
19:21.09 | DocScrutinizer | (my daily business btw) |
19:21.32 | kristianpaul | GNUtoo-desktop: task basically port and make its gui, but disable (ifdef) warm start |
19:21.58 | kristianpaul | make it guis to work* |
19:22.11 | GNUtoo-desktop | ok so it doens't even require gps hardware |
19:22.17 | kristianpaul | nope |
19:22.40 | kristianpaul | neither gps hardware or milkymist it self |
19:22.52 | GNUtoo-desktop | ok |
19:23.08 | GNUtoo-desktop | I'll contact you if I've the time to work on it |
19:23.33 | DocScrutinizer | aaaah, some detail about hsi that now hits my mind: |
19:24.38 | DocScrutinizer | 2 wires. one is either 1 or 0 depending on next byte to transmit. the other one will toggle each time a second 0 (or 1) has to be transmitted |
19:25.00 | GNUtoo-desktop | ok |
19:25.19 | GNUtoo-desktop | I guess that's for error correction |
19:26.23 | DocScrutinizer | that's for sync transmission of serial data, with embedded clock and minimized transitions on lines (there will a transition on only one of both lines for each bit transmitted) |
19:27.39 | GNUtoo-desktop | ok |
19:28.44 | DocScrutinizer | let's call the lines A and B, and the byte to transmit is 01110010: the edge is on line like that: 0A 1A 1B 1B 0A 0B 1A 0A |
19:29.35 | DocScrutinizer | where leading 0/1 means bit to transmit, NOT level on lines A/B |
19:30.34 | DocScrutinizer | for A however it actually means level on line as well, while 0B and 1B both only mean there's a transition on B, either low->high or high->low |
19:32.23 | DocScrutinizer | badically A is a serial UART output, while B is the UART bit clock XOR A |
19:32.36 | DocScrutinizer | basically* |
19:34.55 | DocScrutinizer | you don't need to know any of that |
19:35.40 | DocScrutinizer | way more interesting is the configurble header and dataword length |
19:36.44 | GNUtoo-netbook | ok |
19:36.48 | DocScrutinizer | I think the hw defines a header 0..3bits for logical channel, plus anything from 8 to 32bits for data per packet |
19:37.41 | DocScrutinizer | and you got dedicated hw for each of the up to 8 logical channels |
19:38.48 | DocScrutinizer | 1.2 defines some burst mode iirc |
19:39.04 | DocScrutinizer | "our" (STE) hw doesn't support that |
19:39.13 | GNUtoo-netbook | ok |
19:49.59 | DocScrutinizer | actually claiming I'd work for LTE is a joke. My position is consultant LLSW Datacom |
19:50.11 | DocScrutinizer | LLSW = lowlevel sw |
19:54.24 | GNUtoo-netbook | datacom? |
19:54.52 | GNUtoo-netbook | data communication or the name of a company or something else? |
19:56.25 | DocScrutinizer | data communication aka data protocols, aka monkey for hw interfaces |
19:56.56 | DocScrutinizer | that's how I had to look into HSI some 3 months ago |
20:00.03 | GNUtoo-netbook | ok |
20:00.20 | DocScrutinizer | actually, it's >> Joerg Reisenweber, Consultant Software Design Data Protocols, is Industrie Software GmbH on behalf of ST-Ericsson AT GmbH, Nuremberg << |
20:00.35 | GNUtoo-netbook | ok |
20:24.12 | *** join/#qi-hardware Ayla (~paul@208.131.123.78.rev.sfr.net) |
21:06.39 | *** join/#qi-hardware jurting (~jommpa90_@c213-100-125-198.cust.tele2.se) |
21:26.14 | *** join/#qi-hardware pabs3 (~pabs@d122-111-245-94.per801.wa.optusnet.com.au) |
21:28.32 | DocScrutinizer | whitequark: what's been the name of that gsm-module? |
21:52.23 | *** join/#qi-hardware paroneayea (~user@fsf/member/paroneayea) |
22:13.02 | *** join/#qi-hardware paroneayea (~user@fsf/member/paroneayea) |
22:37.38 | *** join/#qi-hardware wej (~j@95.211.10.3) |
23:49.19 | qi-bot | [commit] Maarten ter Huurne: MIPS: JZ4740: Correct clock gate bit for DMA controller (jz-3.3) http://qi-hw.com/p/qi-kernel/bf21f89 |