<wpwrak>
rafa: (kernel) i think 1-2 mminutes is a good time. no need to exaggerate :)
<rafa>
wpwrak: 120 seconds are few for me.. ;)Â Â it is a good time to heat water to drink mate
<rafa>
wpwrak: btw... I was playing hexen and heretic... wow!..  these are great for nn.. still if I just played the shareware levels.. No idea where to download levels made by users yet.. But the sound.. man.. it is amazing to play..
<rafa>
(I am building games for the trip :D )
<wpwrak>
you'll keep the whole plane awake ;-)
<rafa>
I already built around 10 great games.. I do not want to sleep now :)
<rafa>
hexec would be great for nanowar.. you feel the metal at that game hehe
<rafa>
hexen
<wpwrak>
never played it. should be fun to discover then :)
<rafa>
hexen is very similar to doom.. but there are more effects and the player is more like medieval ages
<rafa>
effects: for example thunders or storms
<wpwrak>
ah, nice. so it's outdoors ?
<rafa>
in fact I guess that hexen, heretic and doom share a lot of code.. when I built those the building output were similar between them :)
<wpwrak>
:-)
<nebajoth>
they do
<nebajoth>
share code
<nebajoth>
I remember hexen
<nebajoth>
mmmm
<rafa>
nebajoth: do you want to play on nn? :)
<nebajoth>
hmm
<nebajoth>
I dunno
<nebajoth>
means installing graphical environment :P
<rafa>
nebajoth: no sure.. it uses sdl and sdl works on fb
<roh>
everything seems to work, it just doesnt enumerate.. /me puzzled. now i really could use a _real_ scope with some usb debug feats
<wpwrak>
roh: any ~100 MHz scope with deep memory will do. i debugged USB enumeration with mine.
<roh>
deep memory ;)
<wpwrak>
500 kSa :)
<wpwrak>
not that it's complain if a genie popped out of a bottle and offered me something with a few more MHz and more samples, though :)
<wpwrak>
s/it's/I'd/
<wolfspraul>
rafa: what is the launcher app you use in Jlime?
<rafa>
wolfspraul: we do not use launcher
<rafa>
wolfspraul: jlime use matchbox desktop and matchbox window manager.. Matchbox uses freedesktop.org spec
<rafa>
So you can add desktop entries under /usr/applications/*.desktop
<rafa>
wolfspraul: those would be launchers I guess. Well, you define *.desktop files like in gnome.. or others freedesktop thing and it is showed on desktop.
<wolfspraul>
ah OK
<wpwrak>
wolfspraul: btw, one problem of ben-wpan for which I don't have a good idea for a solution yet is how to connect it to a ben. particularly if placing the board next to the display
<wpwrak>
wolfspraul: the signals i can grab at the LCM aren't enough. i need at least three more: reset, chip select, and interrupt. there's one more signal used for sleep mode that may also be good to have. if i take the clock from the ben, saving the crystal, that would be one more signal. so, 3-5 signals to route from the main pcb to the display.
<wpwrak>
wolfspraul: ah, one more: i also need MISO. the LCM only has MOSI and SCLK. so that's 4-6 signals.
<wolfspraul>
no problem if we integrate this we have an entirely new layout anyway :-)
<wpwrak>
pity that all the lines of the lcm's fpc are connected. there are many GND and +V3.3 lines. probably half the number would do nicely
<wolfspraul>
is the avt2 I sent you any good?
<wpwrak>
i haven't played with it yet
<wolfspraul>
so basically my plan on the small run side looks like this:
<wpwrak>
small run of what ?
<wolfspraul>
I hope we can continue hacking projects with the existing Ben, AVT2 or SIE boards, until we have a bigger package to go for another NanoNote prototype run
<wolfspraul>
small run anything :-
<wolfspraul>
:-)
<wolfspraul>
so of course we could make another avt2, and bring out everything you need, bit a small run is quite expensive, both in case and because we can only do so many small runs per year
<wolfspraul>
gee, typos
<wolfspraul>
"but a small run is quite expensive, both in cash and ..."
<wolfspraul>
so I am waiting either for a usable 4760, or do the milkymist nanonote right away, but it's a bit too early for that...
<wolfspraul>
which means ... to find your wires I hope we can get it going on a Ben, or AVT2, or SIE
<wpwrak>
MM NN would be fun. should make the headlines as the craziest project of the year ;-)
<wolfspraul>
what do you think?
<wolfspraul>
oh we will definitely do it :-)
<wolfspraul>
in fact we are working on it already, aren't we? :-)
<wpwrak>
i can prototype on the ben, no problem. that's the least of my worries :)
<wolfspraul>
ok but you said you don't know how to hookup ben-wpan?
<wpwrak>
what's harder is to test it in "real life" conditions. because i need to solve the cabling problem for that. well, at least to the point that they only need resoldering once a week or so :)
<wpwrak>
hook it up in a "real life" way
<wpwrak>
just somehow connecting it to the pcb is easy. putting it in the display such that it stays connected when opening/closing the device several times is difficult.
<wpwrak>
what i also don't have is the fancy rf gear for testing spectrum leakage an all that. i think (*) i more or less check if something horrible happens at the first harmonic, but that's about all. (*) still need to get my usrp going.
<wolfspraul>
hmm
<wolfspraul>
integrating it into the top side is indeed tough, didn't know you want to do that now already
<wpwrak>
rf properties are basically a function of circuit design, circuit implementation (including soldering), antenna design, and - you'll like that one - properties of the device this thing operates in. so there are many things you can only find out by measuring
<wpwrak>
basically the y-size of the antenna is what we have. the ground below the antenna can go under the lcm without problems (well, i also hope without rf problems)
<wpwrak>
let me take a picture of the actual circuit ...
<wolfspraul>
oh wooow! that's your own PCB antenna there?
<wolfspraul>
etched in home-made acid?
<wpwrak>
yup :) btw, it has about +3 dB, according to TI's documentation ;-)
<wolfspraul>
you are getting scary
<wolfspraul>
if Al Qaida has any problems in recruitment, someone should point them to you...
<wolfspraul>
the new 'base' is in Buenos Aires. Asado Qaida.
<nebajoth>
pish posh
<nebajoth>
you should know better than to say the words "al qaida" in an irc channel
<nebajoth>
ECHELON HIRE ME
<wolfspraul>
ok then let me rephrase "Werner's PCB antenna is cool"
<wolfspraul>
self-censored
<nebajoth>
:D
<nebajoth>
can't we do a 4740-based small run with usb host?
<wolfspraul>
that was avt2
<wolfspraul>
avt2 had a 4720 which is the same die as 4740
<nebajoth>
I'm half asleep, watching my daughter :P
<wolfspraul>
a small run doesnt' have to be profitable
<wpwrak>
left chip is the usb mcu. right chip is the transceiver. little stuff right of the transceiver is balun and filter. it should be possible to replace them all with a single chip about the size of two of these small critters (they're 0402)
<wolfspraul>
but a 'real' run does, and I don't want to increase prices of the NanoNote, I want to decrease prices
<wolfspraul>
so the 'small runs' will dance around for a while, until we find a package that we can successfully mass-produce
<wpwrak>
everything left of the pads with the colorful little wires would come off then putting the board into the ben
<nebajoth>
yes alright
<wolfspraul>
so what's next?
<wpwrak>
one problem is the crystal. there's no room for it in the ben, if putting the board into the display. but i can probably slave the board off a clock from the cpu. the 12 MHz crystal is almost good enough for this. sufficient for experimental use. for a real product, it may have to be replaced with a slightly more accurate one. shouldn't be a big deal.
<wpwrak>
next in my testing is to enable the receiver and see if the analog regulator comes up too.
<wolfspraul>
we could make a special PCB behind the LCM, then use the current FPC that goes through the hinge to route the wires to the main PCB. but we would then probably also need changes there, so I doubt it will help much.
<wpwrak>
after that, the fun part starts: the RF side. my plan is to first implement a spectrum scan, do the same on the usrp, and compare the two.
<wpwrak>
particularly when generating background noise form my wlan, they both should see the same pattern
<wpwrak>
s/form/from/
<wpwrak>
the lcm's pcb has very large ground areas. i can just cut out these
<wpwrak>
(well, some of them. don't need all that much space)
<wpwrak>
the fpc would need a change on both sides, yes
<wpwrak>
maybe i can cut the traces leading to the fpc connectors. haven't examined that yet.
<wpwrak>
the transceiver's reset behaviour is interesting. i was trying to read the chip's IDs. the data sheet says no explicit reset is needed. plus, spi is self-synchronizing (with nSS) and reading the ids is about the simplest thing one can do anyway. so i expected this to work.
<wpwrak>
instead, i just got random garbage. pretty much all of friday went into trying to find the mysterious bug in my spi implementation. the general shape of the signals was okay. all the bits made sense. signal timing as well, signal shape too. and so on.
<wpwrak>
in the end, it turned out that i do have to reset the chip before it would return anything sensible. argl.
<wolfspraul>
oh well
<wpwrak>
of course this means that having a line for the reset signal is mandatory
<wpwrak>
the one signal that's kind of optional controls transmit timing and sleep mode. we probably don't care about the former, but the latter is interesting: the chip goes from 1.5 mA to 20 nA in sleep mode.
<wolfspraul>
you think the aread around the lcm (on the top side) is the right place to attempt RF, right?
<wolfspraul>
there are orcad files of the behind-lcm pcb flying around somewhere, as a first step I could try to turn this into proper kicad files.
<wolfspraul>
one problem with converting GND lines in the FPC is that they are probably there for a reason, to reduce noise or whatever
<wpwrak>
i think the lcm area is promising, yes. it would be far from parts with complex radiation patterns, next to a big chunk of ground, and it's likely to be uncovered by hands and such
<wpwrak>
some of the ground lines may be there for noise. but most seem to be there just for current. i don't think we need quite so many. even one of each (gnd and +V3.3) is probably enough. two of each should be plenty.
<wpwrak>
then there are two more ground lines that may be there for noise. and i'm adding another even noisier line, the 16 MHz clock. i think the lcm clock is much slower than this. maybe 8-10 MHz
<wpwrak>
actually, i could measure it ... let's see ...
<rafa>
wolfspraul: BTW, beta3 is almost there ;)
<wpwrak>
ah no, it's 21 MHz
<wolfspraul>
great!
<wpwrak>
so the clock i may add will be a little less offensive than the one that's already there ;-)
<wpwrak>
in any case, i know that adam likes this kind of signals. we already had great fun with that sort of thing with the gta03 camera clock :)
<wpwrak>
wolfspraul: how is the EE design of the bens done ? is it split between carlos and adam ?
<wolfspraul>
basically yes, although keep in mind be bought the design so we started with a lot of 'known good' data
<wolfspraul>
s/be/we/
<wolfspraul>
but then avt2, sie - yes, Carlos & Adam
<wpwrak>
okay, good. does carlos have resources to analyze RF characteristics ? such as the sort you need for FCC compliance
<nebajoth>
is it mipsel or mipsn32?
<wpwrak>
nebajoth: openwrt toolchain calls it mipsel-*
<nebajoth>
yeah
<nebajoth>
but the kernel config for 2.6.36 seems to think its mipsn32
<wpwrak>
where is a -march=xburst when you need it to end all doubts ;)
<wolfspraul>
wpwrak: I doubt it. Although it's a university someone else might have it.
<wpwrak>
wolfspraul: that's what i was hoping for :) okay, something to find out
<wpwrak>
rafa: we should find out if there are any people at universities in buenos aires who could be interested in such things as well
<wpwrak>
nice. i get non-constant rssi values. this is promising :)
<wpwrak>
x is the frequency, in MHz. y are the seconds since the start, z is the received energy in dBm. at t=~6s i started a lot of traffic on my wlan, at channel 10 (2457 MHz)
<wpwrak>
in other words, i get something out of the ether
<nebajoth>
:D
<wpwrak>
updated the image
<nebajoth>
nice
<nebajoth>
labels
<nebajoth>
I kind of want this kind of display live-updated on my wall
<nebajoth>
or wristwatch
<wpwrak>
shouldn't be too hard. Gtk is your friend :)
<qi-bot>
[commit] Werner Almesberger: Additions to bookshelf: AT86RF230 programmer's guide and 2.4 GHz regulations. http://qi-hw.com/p/ben-wpan/d8c71a3
<wpwrak>
wolfspraul: you'll like the last one :)
<wolfspraul>
regulations! my everyday bedtime reading
<wolfspraul>
fantastic Werner, thanks a lot!
<wpwrak>
but i can't praise ti's wonderful collection of documentation often enough. they really have everything you need.
<wpwrak>
wolfspraul: i thought it was your bathroom reading, with dual use ;-)
<wolfspraul>
can't get enough regulatory texts actually
<wolfspraul>
the grammar, style, vocabulary - an ecstatic pleasure
<wpwrak>
hehe :) well, this one is an overview, so it's pretty decently written. the originals are probably a different story.
<wolfspraul>
wpwrak: I need your help deciphering the extensions of some KiCad files, and what should be committed into the repository, what is intermediate/temporary
<wolfspraul>
it's about the JTAG/serial reflashing cable for Milkymist One, a very simple little PCB
<wolfspraul>
some files are already committed, but neither me nor Yanjun Luo are really sure whether it's the right set
<wpwrak>
the .dcm is definitely useless. has no content.
<wolfspraul>
leaves 14
<wolfspraul>
ok, 13
<wpwrak>
for the netlist, make sure you have a copy. just in case.
<wolfspraul>
he well
<wolfspraul>
that's just what I am trying to find out
<wolfspraul>
also for the .dcm - if it's empty here that doesn't mean it won't contain original content alter
<wolfspraul>
later
<wolfspraul>
so it's important whether it's committed to git or in the ignore list :-)
<wpwrak>
naw, it's a cache auto-generated by kicad
<wolfspraul>
ok
<wolfspraul>
so we are down to 13
<wpwrak>
the netlist should have no original content, but i'd do a test to verify that
<wpwrak>
you killed $savepcb.* and *.000 ?
<wolfspraul>
maybe this is documented in the kicad wiki/documentation somewhere?
<wolfspraul>
I killed?
<wolfspraul>
no
<wpwrak>
then do it :) 3 less
<wolfspraul>
ok, let me move all probably-non-original files into a subfolder
<wpwrak>
the netlist is generated by eeschema and cvpcb updates the netlist when you assign footprints to components. in this case, it should save them also in a .cmp file, though.
<wpwrak>
footprints can also be directly in the schematics, so you don't need to run cvpcb
<wolfspraul>
wow, so some .lib have original content, some not?
<wpwrak>
nice, eh ? ;-)
<wolfspraul>
that's not nice even if you hate extensions
<wolfspraul>
down to 8
<wpwrak>
what kicad does is that it copies the libraries it uses into a cache. if you use only stable "standard" libs and supply the non-standard libs with the reset of the project, the cache is redundant.
<wpwrak>
now you have all you should need. lemme check the .pro ...
<wolfspraul>
so the .lib files need to be committed
<wolfspraul>
there's also jtag-uart.mdc and jtag-uart.mod
<wolfspraul>
both are original?
<wpwrak>
the .pro should be edited. has a long absolute path name. i also don't see the .lib files mentioned in there
<wpwrak>
best to clean out all the junk (LibName30=valves, LibName22=digital-audio, etc.)
<wolfspraul>
ok, will fix the absolute path
<wpwrak>
then add the *.lib files. as  ./crystal_smd  etc.
<wolfspraul>
well there is a crystal_smd.lib, and if I grep -i crystal_smd, I do find some references to that name here and there
<wolfspraul>
in usb_jtag.brd and usb_jtag.sch
<wpwrak>
yes. it should also be in the profile. not sure if eeschema can even display it if it's not there.
<wolfspraul>
maybe it just suffixes a '.lib' and tries to open the file :-)
<wolfspraul>
I don't know
<wolfspraul>
so you think what's remaining there now are the files with original content?
<wolfspraul>
they should go into revision control, the rest can be ignored
<wpwrak>
and i'm not sure why it's not there in the first place. maybe there's yet another strange workflow that lets you pick libs up without adding them to the list :)
<wpwrak>
yes
<wolfspraul>
what is in jtag_uart.mdc and jtag_uart.mod
<wolfspraul>
mdc is quite short, mod rather long
<wpwrak>
.mod is a footprint, .mdc is documentation
<viric>
Hello all
<viric>
I have a linux question...
<wpwrak>
wolfspraul: ah, for once there is actually something in the documentation ;-)
<wolfspraul>
ok but unlike .dcm, the .mdc documentation is original?
<viric>
If I wanted to do an ubifs image, so it had the /dev/... contents, how could I do that on a linux without being root?
<wpwrak>
wolfspraul: yup
<wolfspraul>
ok
<wolfspraul>
so at least in terms of original vs. intermediate/generated, the split is rather clean
<wolfspraul>
we have all generated files in non-original, and the ones with original data remain, 8 out of 22 here
<wolfspraul>
true?
<wpwrak>
wolfspraul: yup. it's manageable :)
<wolfspraul>
wpwrak: thanks a lot! very helpful
<viric>
It's a similar problem with making an initramfs image. The only tool I know is one comming with the linux kernel, that allows putting device ndoes into a cpio archive according to a description, not only through the nodes being in the filesystem.
<wolfspraul>
I will try to write this up into an email to the list, good that we have irclogs...
<wpwrak>
wolfspraul: depending on the kicad version, things change a little. i.e., it sometimes it complains if certain files are missing (but works anyway)
<viric>
do you talk about the 3d scans?
<wpwrak>
viric: no, about kicad. that's EDA (schematics, layout, etc.)
<viric>
ah ok
<wolfspraul>
wpwrak: ok, but I would like to commit only original data
<wolfspraul>
then it's a bug in KiCad :-)
<viric>
ok. schematics, layout, for what? the nanonote?
<wolfspraul>
they must have heard about the idea of separating original data from intermediates/generated data
<wpwrak>
one more scan shuold end today. quite a bastard been running for something like 3 or 4 days now. not sure why it takes *that* long.
<wpwrak>
viric: ultimately, that one too :) for now it's assorted smaller projects
<viric>
ah ok
<wolfspraul>
although it seems quite brave to use the same extension for both original and cached data, first time ever I see something like this
<wpwrak>
viric: e.g., a JTAG dongle for Xue, or ben-wpan
<viric>
On Linux I only used Eagle... it looks quite similar
<viric>
Ah, very good.
<wolfspraul>
so maybe they don't care so much about that distinction...
<viric>
is your chosing of 'kicad' the result of exploring what kind of free available software there is for that?
<wpwrak>
hmm. i should make a c8051f326 programmer based on a c8051f326. then i don't need to abuse a freerunner for bootstrapping all my little usb toys :)
<wpwrak>
wolfspraul: the whole library system is ugly. there's talk about improving it, but the powers that be are slowing things down :(
<wpwrak>
viric: yup
<viric>
I'll try this kicad then :)
<wpwrak>
viric: there are basically two: gEDA and kiCad. i used gEDA once but it was quite a pain. e.g., you had to put meta-data into invisible fields in the schematics and define your schematics symbols in an m4-based script. not sure if things are still like that.
<wpwrak>
then i stumbled upon kicad. never looked back :)
<viric>
great
<wpwrak>
the main problem with kicad is that it doesn't have an autorouter. you can use an external non-Free but gratis push router, though.
<viric>
ahh.
<viric>
there may be people working on that though
<wpwrak>
well, it does have an autorouter and even an autoplacer, but they're both useless.
<viric>
don't have the geda people one?
<viric>
:) ok
<wpwrak>
they have a fairly nice autorouter. not sure if they have a push router these days.
<viric>
a 'push router'?
<wpwrak>
push router = you route manually but the program moves conflicting traces out of the way
<wolfspraul>
viric: same here, definitely really trying hard to work with KiCad now.
<wolfspraul>
I just follow Werner's lead on choosing that tool.
<viric>
wolfspraul: same for Eagle you mean?
<wolfspraul>
Eagle is not free softare.
<wolfspraul>
software
<viric>
I know I know
<wpwrak>
so it's the best of both worlds - you don't need to waste time on those small details and you're still in control of how things are routed. (autorouters are essentially impossible for anything but trivial layouts)
<wolfspraul>
viric: sorry then I don't understand you. What is your question about Eagle?
<viric>
I had no question. :)
<viric>
wolfspraul: well, my question was about your statement of "same here". I lost what could you be answering, with "same here". And I asked if that "same here" meant "same we used Eagle here"
<wpwrak>
wolfspraul: btw, just how much do you dislike TI ? they have an 802.15.4 chip that's similar to atmel's but has more buffer space. atmel's has only space for one frame, so you have to be very quick to download a received frame or to disable the receiver. otherwise, new packets can overwrite old ones. not sure if this will be a problem, but it might.
<wpwrak>
wolfspraul: so it may make sense to also evaluate TI's chip.
<viric>
is 802.15.4 (wpan?) something like bluetooth?
<wpwrak>
sort of, yes. it's for example the basis of ZigBee.
<wpwrak>
apparently, one couldn't write a Free ZigBee stack, though.
<viric>
ah
<wpwrak>
but there are other nice protocols one can layer on top :)
<viric>
like ip?
<wpwrak>
e.g., 6LoWPAN, IPv6
<viric>
ok
<viric>
are you planning an 'additional small board' to give network to the nanonote?
<viric>
OTH, I thought that through the serial port it could be easy to handle a gprs modem through that, right?
<wpwrak>
no IPv4, though. i think you'd get a public stoning if you were to suggest to start an IPv4 project at IETF these days ;-)
<wpwrak>
that's the transceiver (a bit too large to fit into the ben but that can probably be fixed) plus an USB interface. the latter for development.
<viric>
wow, it looks great.
<viric>
is that 'low cost'?
<wpwrak>
quite, yes. about USD 2-3 for the whole thing we'd need for the ben. maybe twice that for USB.
<viric>
wow
<viric>
sounds very nice
<viric>
and a wpan transceiver for a PC costs around that?
<wpwrak>
you can find cheaper RF chips, but then you're either in bands with hostile regulations or they implement some "vendor standard", incompatible with everything else
<viric>
I think anyone who bought a nanonote can afford this wpan thing
<wpwrak>
dunno what commercially available WPAN/ZigBee dongles cost
<wpwrak>
it's not an easy upgrade. it will have to be engineered into a new ben.
<bartbes>
:(
<wpwrak>
adding it to an existing ben may be unreliable and you'd have to cut the PCB in the display and route some new wires to the LCM. not very friendly.
<viric>
it's a pity the Ben did not come with an 'extension port', with exposed pins
<wpwrak>
installing an IDBG is a kid's play in comparison :)
<viric>
ok, understood :)
<wpwrak>
i'm not sure an extension port would help. chances are it's the wrong shape, at the wong place (rf likes to be undisturbed), and maybe even with the wrong signals.
<viric>
all rf could be outside the ben
<viric>
and something connected to that ext port.
<wpwrak>
ugly
<viric>
:)
<wpwrak>
and the sooner or later, you'll break it off :)
<viric>
right
<viric>
do you have the schemas in kicad for that?
<wpwrak>
now, let's see if my hands are steady enough to solder after yesterday's drinking ...
<viric>
you must be a good soldering, seeing that small (smd?) chips soldered
<wpwrak>
it's not so difficult. i find smd easier than through-hole.
<viric>
sure
<wpwrak>
the main difficulty on this board is that there isn't a lot of space between components. spacing is still wide for industrial production, but with a soldering iron, it's about as dense as possible.
<viric>
wolfspraul: is it you doing those long-time 3D scans?
<viric>
wpwrak: clear
<wpwrak>
well, "industrial production" ... a toaster oven is sufficient :) but i like the results of manual soldering better.
<viric>
I never used other than manual soldering
<wpwrak>
for proper reflow soldering (with an oven), i would need to print stencils for the solder paste too. haven't gotten into that yet.
<wpwrak>
i'm doing the 3D scans. well, my little cnc mill that ever sleeps is :)
<viric>
ah ok!
<viric>
I should be in charge of doing 3D scans with laser triangulation
<viric>
But we are very busy at the office these days, and I think I will not get into scanning until the late september
<viric>
(last week of september)
<viric>
wpwrak: I don't think I'll get results as accurate as you (in 0,1x0,1µm scans), but I hope we can get something close to that.
<viric>
at least I should be able to scan faster (30s per scan at that resolution)
<wpwrak>
my scans are 100 um x 100 um, not 0.1 um. that would be nice, though :)
<wpwrak>
(30s) i envy that speed :)
<viric>
sorry :)
<viric>
I meant 0,1x0,1mm :)
<wpwrak>
with those endless scan jobs, i'm extremely nervous about brown-outs. they happen here from time to time. and of course, they kill the entire scan. the crappy scan software doesn't to intermediate saves :-(
<viric>
oh.
<viric>
maybe you can leave your CNC relax...
<viric>
if you can wait for my results
<wpwrak>
what accuracy do you expect to get ?
<viric>
In general we only do algorithms, and we lack good scanning hardware :) Our customers get the expensive hardware and use our algorithms
<viric>
I hope I will be in the 0,1x0,1x0,1µm range.
<wpwrak>
(customers) ah. pity. it's always nice to have good toys :)
<wpwrak>
0.1 um or 0.1 mm ?
<viric>
grrrr
<viric>
100µm^3
<viric>
:)
<wpwrak>
;-)
<viric>
we could scan at more accuracy
<wpwrak>
i get 100 um on X/Y and 25 um on Z.
<viric>
but I think we will be around that error for the time of integrating multiple views.
<wpwrak>
i guess the better the scans, the better the merged views, no ?
<viric>
but for very flat objects, it is very difficult to get common surfaces between the views.
<viric>
flat objects defeat the usual aligning technology.
<viric>
this could take both top and bottom views of a flat surface at the same time
<viric>
of a flat object I mean :)
<wpwrak>
well, i'll just keep my scans going. when you get to make yours, we can compare. if the equipment is good enough, then you'll obsolete my scans within a few minutes :)
<viric>
I hope to do so ;)
<viric>
Oh, I got kicad built. Time to experiment
<wpwrak>
my scans don't have common surfaces. (i could do them, but i'm not sure they'd come out great, with items mounted at an angle) so i have to measure the thickness of opposing surfaces.
<wpwrak>
(at 1 um resolution. micrometers still beat scanners ;-)
<viric>
wpwrak: right, that's a big source of error I see with one scan at every position.
<viric>
ah, let's see what you get then :)
<wolfspraul>
wpwrak: I have no general problem with TI, that would be pretty irrational and stupid imo.
<wolfspraul>
if they have (bought) chips that perform well, sure let's take a look.
<wolfspraul>
which one of the two is cheaper - atmel or ti?
<wpwrak>
wolfspraul: atmel is a bit cheaper. lemme look up the digi-key prices ...
<wpwrak>
atmel: 2.2724@100, 2.09352@1000, 2.15384@5000(reel). i never quite understand why TR has a higher per unit cost than custom quantities plus reeling.
<wpwrak>
ti: 4.73@100, 2.59350@3000(reel)
<wpwrak>
ti themselves list their chip as 1.95@1k. so there seems to be some room for negotiations ;-)
<wpwrak>
the atmel is at86rf230. the ti is cc2520.
<dmarschal>
hi!
<dmarschal>
can someone help me how to run native gcc on my nn?
<dmarschal>
i tried it under jlime but the keyboard is acting weird -- i can't type numbers
<dmarschal>
i checked all gcc related stuff in the xburst repo. i've found nothing to 'opkg install'
<dmarschal>
any advices?
<bartbes>
have you run opkg update before install?
<dmarschal>
i don't have a working inet connection
<dmarschal>
i'm running the 8-15 openwrt userland
<bartbes>
well, then you either need to build the package, or download it on your computer
<bartbes>
I guess
<bartbes>
I'm no expect, maybe someone else knows a different way
<wolfspraul>
wpwrak: ok got it. seems they are similar, although it's very hard to tell from digikey.
<dmarschal>
bartbes: i hope so. i tried to cross compile gcc and transfer to my NN but it doesn't work. missing deps maybe.
<wolfspraul>
you should feel free, really. if the TI chip is better it's better.
<wolfspraul>
I'm a pragmatist, both are proprietary ICs in the end...
<wolfspraul>
I have heard many good things about the ti cc series
<dmarschal>
so everyone is cross compiling even simple 'hello.c' applications? i think a 300mhz device is powerful enough to compile \
<dmarschal>
even graphics applications (not x11)
<wpwrak>
wolfspraul: they're quite similar in terms of general characteristics and the number of external components. ti's has more buffers and an AES engine. haven't looked at the programming model yet.
<wpwrak>
wolfspraul: kewl. adding cc2520 to my digi-key shopping list :)
<wolfspraul>
comparisons are always good, real comparisons, not spec comparisons :-)
<wolfspraul>
marketing people are quite good, maybe terrific, when it comes to what to talk about and what not to talk about...
<viric>
dmarschal: most problems compiling come from the memory available, not the cpu speed
<dmarschal>
viric: thank you. i was able to compile sdl codes (even mplayer) on my zipit z2. it's almost the same device.
<wpwrak>
wolfspraul: indeed. gta01 gps, that wlan module by microchip, etc. ;-)
<viric>
dmarschal: maybe with an old gcc? With swap? or how?
<dmarschal>
viric: checking gcc version...
<dmarschal>
viric: gcc version is 4.3.3. i set up 130mb swap space.
<dmarschal>
it compiles sdl examples (from sdl.org) in a few seconds. so I assume the NN is able to do more.
<dmarschal>
sorry, from libsdl.org
<viric>
ahh with swap. that's cheating :)
<dmarschal>
but it worked (-:
<dmarschal>
zipit has thumb cracker keyboard that's why i need my nn to work as it should
<viric>
I could run gcc in the nanonote fine
<viric>
cross-built gcc.
<viric>
what did you say about the internet connection?
<dmarschal>
i don't have a working connection on my nn
<viric>
why not?
<dmarschal>
the host computer drops connection every 3-5 secs.
<dmarschal>
because of power saving
<dmarschal>
it's a fujistu u1010 with ubuntu
<viric>
drops the usb?
<bartbes>
are you sure it isn't nm-applet?
<dmarschal>
i have to check. thanks for the idea. by the way ubuntu on the u1010 is not perfect
<dmarschal>
nightmare starts when i want to set up a connection between those 2 computers.
<viric>
Is anyone happy with the recording quality of the microphone?
<viric>
of the nn
<viric>
I've noticed that not all 'arecord' options work fine; does anybody know what should I tell to arecord to get the nicest quality?
<kyak>
viric: for some reason, arecord is not working for me at all
<kyak>
at least, the wav file i get is not played back
<viric>
I could record, but I got 'normal' results only at 8000Hz
<viric>
ah, strange. it works for me.
<kyak>
what is your command line?
<viric>
arecord a.wav
<viric>
aplay a.wav
<viric>
that gives unsigned 8 bit, 8000Hz mono
<viric>
I hear some 'crack crack' in the middle of the recording though
<viric>
you have to set up the capturing with alsamixer maybe. Did you do that?
<kyak>
no, i don't get anything
<kyak>
yes, i tried playing with alsamixer
<kyak>
alsamixer - F4 - Mic set to 100 %
<viric>
do you have CAPTURE enabled there?
<viric>
(space key)
<kyak>
in "Capture" section of alsamixer there are Line, Mic, Capture
<viric>
hm
<kyak>
Mic and Capture are set to 100, Line is enabled (with space key)
<viric>
I have 'master, line, mic'
<kyak>
this is in "F5"
<kyak>
i.e., "All" section
<viric>
mm no. in f5.
<kyak>
oh!
<kyak>
i got it!!
<viric>
in f4 I meant
<kyak>
though Mic was set to 100 %, it wasn't enabled (space key)
<kyak>
viric: thanks, now it works :)
<viric>
do you get good sound?
<viric>
be welcome
<kyak>
no, not very good
<kyak>
the sound level is low, there are "cracks"
<kyak>
speaking in a normal voice when sitting at the table where Ben is is recorded almost unhearable
<viric>
I guess it's not easy to configure it
<kyak>
speaking directly in a mic produces mor or less suitable results
<kyak>
still a lot of background noise
<kyak>
viric: arecord -f cd a.wav produces better results i think
<kyak>
the sound is louder
<viric>
really?
<kyak>
yes..
<viric>
ah yes. still cracks though
<kyak>
yeah
<viric>
the mic volume seems to do little
<viric>
but the 'master' works fine
<kyak>
if you don't touch Ben during recording (Ben is on a table), there are not cracks
<viric>
really?
<kyak>
i think these cracks are from Ben internal's somewhere :)
<kyak>
when you touch buttons, etc
<viric>
mm I hear cracks though. less, but I hear.
<kyak>
i hear background noise
<rafa>
wolfspraul: BTW, any suggestion for the current GUI?
<wolfspraul>
you mean in Jlime?
<rafa>
wolfspraul: yeah.. you asked me about the launcher and I was thinking that you could have some suggestions.. ;)
<wolfspraul>
he, no. I need to try first.
<wolfspraul>
I was just curious what you are using, because I was thinking about gmenu2x on the OpenWrt side.
<rafa>
wolfspraul: ah okey. You should try ;) We like the chance to run several applications at the same time. Also, we can use many of the nice things written for desktops, like applets, xlibs applications, gtc, etc.. And all the current sdl applications I have seen on openwrt would work, because sdl works on X as well.
<wpwrak>
nice. 2nd board works, too.
<wpwrak>
(just passed the rssi scan)
<kristianpaul>
rafa: whats the package name of your wikipedia hack in jlime?
<rafa>
kristianpaul: no package yet, somebody should do that ;)
<viric>
do you use any wikipedia offline reader in the nano?
<rafa>
kristianpaul: let me check.. I think that it is not uploaded.. I just have upload the files on temp when people were interested to test, and also I needed to give them instructions.
<rafa>
kristianpaul: so if you want.. then just ask me .. (I need to have more time to do the package it seems :) )
<kristianpaul>
rafa: i just want a try now that finally got 2Gb sd
<wolfspraul>
wpwrak: are you ever sleeping?
<rafa>
viric: what do you think about the idea? (of the wiki demo)
<wolfspraul>
I just wrote up our KiCad extension chat from earlier into a mail, will send it now
<rafa>
kristianpaul: okey.. let me a while.. working on beta3 and I need a bit of time to upload the files
<kristianpaul>
rafa: first beta3 i can wait :)
<rafa>
wolfspraul: that guy is lazy
<kristianpaul>
alo i need learn package for openwrt and jlime ... so i may have an excuse :)
<wolfspraul>
I noticed there are many more extensions in KiCad - .pho .erc .drl .cmp, .pos, .rpt, .stf, .fpd, .cache (instead of -cache)
<viric>
rafa: let me watch the video first..
<kristianpaul>
your wiki reader seems too  portable for xburst-openwrt :)
<wolfspraul>
they must be out to get a record for most extensions or so
<viric>
rafa: ah, it's a dictd server serving the wikipedia?
<rafa>
kristianpaul: I did a first shell command to get articles.. so yes... using that you can use it on different GUIS ;)
<rafa>
viric: no.. it is a offline wikireader.. and we did a mydict plugin to use it. NO dictd server
<wolfspraul>
watch how long it takes until the page is completely loaded
<wolfspraul>
I am trying to debug some slowness on that page but strangely are not able to track it down somehow...
<kristianpaul>
wolfspraul: 5 secnds
<kristianpaul>
8 seconds with epiphany
<wolfspraul>
hmm. ok that's about right.
<kristianpaul>
the other measure was in iceweasel
<wolfspraul>
maybe some strangeness on my client's end
<kristianpaul>
my home link is 1Mb/s
<wolfspraul>
it's grabbing lots of stuff from all over the net (naturally being a planet), but my Firefox keeps hanging supposedly at some elements coming from qi-hardware.com, which I don't understand
<wolfspraul>
or maybe it's just misinformation
<kristianpaul>
hmm
<kristianpaul>
iceweasel dint said nothign about
<wolfspraul>
ok thanks for testing, I'll leave it as is right now
<kristianpaul>
sure is great as i can see
<wpwrak>
wolfspraul: can i have another usb id ? :) cb72, for CNTR. no good mapping in "leet" for the N, but the B is conveniently close. it's for a little frequency-counting instrument i'm making for ben-wpan.
<wolfspraul>
sure sure
<wolfspraul>
I just sent you a mail regarding kicad success, process, etc.
<wolfspraul>
or rather to the list
<wolfspraul>
we are definitely making good process on the kicad front, but the process is not well documented, polished, or proven yet
<wolfspraul>
I pasted some links too to illustrate some points, you will see...
<qi-bot>
[commit] Werner Almesberger: cntr/cntr.brd: remove the ground fill. The board got a bit messy with it. http://qi-hw.com/p/ben-wpan/9c2deb8