xiangfu has quit [Remote host closed the connection]
xiangfu has joined #qi-hardware
porchao has quit [Ping timeout: 240 seconds]
porchao has joined #qi-hardware
<nicksydney>
i was really blown away when i peel off the paper..didn't expect it was going to be soo damn good :)
lilvinz has quit [Ping timeout: 240 seconds]
lilvinz has joined #qi-hardware
<nicksydney>
wpwrak_: done the LED wheel...and tinning the track helps a lot in soldering...one thing i notice when soldering the GND the solder does not stick well looks like the heat is "absorbed" by the surrounding copper ?
<nicksydney>
those LED SMD really pack some punch really bright
<wpwrak_>
nicksydney: ah yes, large ground areas do that :) you can a) use more heat, b) heat the PCB from the bottom, c) break down the ground areas to be a bit less conductive.
<wpwrak_>
a) increases the risk of damaging traces. b) is usually messy. the feasibility of c) depends a bit on the situation
<wpwrak_>
(leds) pretty. now christmas can come ;-)
<nicksydney>
wpwrak_: christmas comes late .. better late than never i suppose :)
<nicksydney>
i think i will work off with braking the ground into different section
<nicksydney>
having the tinned track make it easy to solder the component..all i have to do is just place the component in its position and heat up the tinned solder and the component straight way hooked
<DocScrutinizer05>
>>Over-voltage protection USB: Fully integrated up to +28 V (integrated OVP FET)<<
<DocScrutinizer05>
I think Andy's approach of "show me a charger that kills the device and we will consider improving our OVP beyond the 5.5V we got" is not sustainable
<DocScrutinizer05>
even less is the "despite ABS MAX is 5.5V, I haven't seen a device go south on even 9V"
<wpwrak_>
28 V .. nice
<DocScrutinizer05>
btw you *might* want to keep a local copy of this pdf, it seems a pretty rare leak of otherwise highly classified Qualcom docs
<DocScrutinizer05>
the qualcom would be a really nice chip, if it wasn't for this insane NDA stuff and the fact that modem shares RAM with APE
<DocScrutinizer05>
for the average hw manuf or OEM as well as for NSA it at least doesn't matter, for NSA etc it even is the best design they could hope for ;-)
<DocScrutinizer05>
for free hardware it is unbearable
lekernel has quit [Ping timeout: 252 seconds]
neildaemond has quit [Quit: WeeChat 0.4.2]
<nicksydney>
thought 0603 is already small for me to solder...now i'm seeing 0402 ..i'll be damned ..that thing is TINY !
<wpwrak_>
;-)
<wpwrak_>
0402 is just the right size. that is, unless you need to run traces under the chip. that you can't do with 0402.
jekhor_ has quit [Ping timeout: 265 seconds]
kristian1aul has quit [Quit: Reconnecting]
kristianpaul has joined #qi-hardware
kristianpaul has joined #qi-hardware
kristianpaul has quit [Changing host]
<nicksydney>
spoke too soon....there are more
<nicksydney>
01005 (0402 metric): 0.4 mm × 0.2 mm (0.016 in × 0.0079 in). Typical power rating for resistors = 0.031 watt[8]
<nicksydney>
0201 (0603 metric): 0.6 mm × 0.3 mm (0.024 in × 0.012 in). Typical power rating for resistors = 0.05 watt[8]
<nicksydney>
0201 and 01005
<nicksydney>
0.4mm .... speckle :)
<wpwrak_>
yeah, but these are on the nasty end of the range. and they don't give much of a benefit since you need to have some keep-out area around them anyway.
<wpwrak_>
and the keep-out area doesn't shrink. it's basically the size of your soldering iron's tip.
<wpwrak_>
if you go below that, you must use reflow soldering and rework gets complicated
lekernel has joined #qi-hardware
<nicksydney>
i can imagine how complicated that will be... :) ... phones must be using those small packages i assume
* nicksydney
looking around on eBay to buy more resistors and LEDs :)
<DocScrutinizer05>
btw here's another one of those rare docs: http://d-h.st/xJ9
<DocScrutinizer05>
nicksydney: 0201 are kinda standard on phones nowadays, yes
<nicksydney>
DocScrutinizer05: wow ! don't want to mess around with that kind of small package ... it's very tiny..i can go blind soldering it :)
<DocScrutinizer05>
you can't realistically solder that stuff by hand
<DocScrutinizer05>
0402 is nasty for soldering
<DocScrutinizer05>
particularly when your eyes are 50+ years old
<DocScrutinizer05>
I heard 0201 start dissolving completely in high humidity environments ;-P
<nicksydney>
which means phone stop working then :)
<DocScrutinizer05>
even a 0402 can vanish into a parallel dimension when you drop it on your workbench during soldering. It never again shows up, no matter how long you search for it
<DocScrutinizer05>
0201 will simply dive into the solder on your soldering iron tip and you can't get them out of the solder anymore ;-P
<nicksydney>
don't want to imagine it falls on your food and you accidentally ingest it
<larsc>
if you are not careful you might inhale them ;)
<DocScrutinizer05>
yeah, nano particles are considered a potential health hazard X-P
* DocScrutinizer05
eagerly waits for the cold solder, some glue that has very low electrical resistivity and cures during a few minutes or even by gently heating or exposing it to UV
<wpwrak_>
and 01005 is where homeopathy finally meets engineering
<DocScrutinizer05>
if you need, we can pay for a eval board available for that chip
<DocScrutinizer05>
though I'd prefer you to do more than just crtouch10 and we ship one of the Neo900 protos to you ;-)
<larsc>
i can pick one up from Dr. Nics lab
<larsc>
it's not that far away
<DocScrutinizer05>
good!
<DocScrutinizer05>
:-)
<DocScrutinizer05>
we need to ship at least a BSP on debian basis with the devices
<DocScrutinizer05>
such BSP of course needs a working driver for the touch controller, among other stuff
<DocScrutinizer05>
and we pretty much would like to evaluate the hw before we start MP, as well. To avoid issues like jitter etc
<DocScrutinizer05>
might need some LPF on the 4-wire, or sth
<DocScrutinizer05>
only proto eval will tell
<larsc>
I have no idea about how to create a debian SDK
<DocScrutinizer05>
err, that's no witchcraft. Just use a bare bones debian system and provide all the drivers for the particular hardware on the device on top of it
<DocScrutinizer05>
then *maybe* add some *very* basic test UIs for stuff like the crtouch10, that maybe shows the raw output from /dev/ts or whatever
<DocScrutinizer05>
just a comprehensive set of device drivers and as icing on top some testframe that allows to see the hardware operate
<DocScrutinizer05>
for modem that would be an AT minicom-alike interface to control the critter (confort version: do some "macro" to automatically execute a sequence of predefined AT commands to configure the modem according to what we recommend, sth like (8)chat)
<DocScrutinizer05>
plus an audio driver that pipes audio e.g. to and from the wired headset
<DocScrutinizer05>
plus a network device that you can ping or whatever you like to do with it, for the GPRS
<DocScrutinizer05>
we will need some additional tools and stuff, to demonstarte how to do the tight modem monitoring - for TX activity, power-up state, GPS antenna enabled, etc. This boils down to monitoring a few GPIO
<larsc>
right
<DocScrutinizer05>
similar driver+testframe combos will be needed e.g. for the FMRX, FMTX, of course the WLAN/BT (all 4 prolly one module TiWi5)... etc pp
<DocScrutinizer05>
for WLAN the testframe is iwconfig and the driver is wl12xx.ko I'd guess
<DocScrutinizer05>
when for a certain subsystem it's hard to write a testframe but there's already a decent app in e.g. SHR existing, and that app allows to test the hardware in a non-obscure way, then even better
<DocScrutinizer05>
e.g for crtouch10 we won't need a silly testframe like /usr/bin/gnu/od -An -tx1 -w16 /dev/input/ts when there's a drawing app already, that even makes use of pinch/rotate
<DocScrutinizer05>
admittedly a very highly sophisticated one, since the stantum guys wanted to sell their stuff and thus needed something catchy, with a high eyecandy gactor
<DocScrutinizer05>
factor even
<DocScrutinizer05>
doesn't matter when the thing has framebuffer instead of X11
<DocScrutinizer05>
you get the picture
<DocScrutinizer05>
all it needs is a way to start it (via remote shell, or by 27 times pressing the aux button) and it demonstrates the hw capabilities and actual working hardware
<DocScrutinizer05>
when the drivers are already fit for "upstream" then all the better, but that's not mandatory
<DocScrutinizer05>
as long as somebody savvy can create upstream-compliant drivers based on what the BSP demonstrates
<DocScrutinizer05>
in that regard what freescale provided are probably typical low quality BSP driver implementations
<larsc>
that sounds like a list of things that could keep a couple of people busy for a couple of months
<DocScrutinizer05>
I don't think it's that much
<DocScrutinizer05>
modem audio and audio in general is a nasty beast
<DocScrutinizer05>
the rest looks pretty manageable
<larsc>
we'll see
<DocScrutinizer05>
for 95% of the subsystems there already exist complete linux driver implementations
<DocScrutinizer05>
the TiWi5 for example is based on the tw1273 which is used in maemo Harmatten
<larsc>
tell me when you have a prototype ready and I'll see what I can do
<DocScrutinizer05>
sure
<DocScrutinizer05>
:-)
wej has joined #qi-hardware
<DocScrutinizer05>
btw when you'd be interested to majke some real money, I have a "daughter project" in mind, which is sth like "port maemo fremantle and make it security hardened" means set up a few existing security enhancements on top of it, like cryptfs on rootfs, VPN, dunno what else we'd need. Then we would sell this distro with the device and you could tell what you want to ask for it from customers. I.E. we offer devices with "fremantle
<DocScrutinizer05>
security anti-NSA edition" preinstalled and ask a 25 or 50 EUR for the software installation, which goes mostly to you
<DocScrutinizer05>
I have prospects for a few 1000 devices like that, just here in Germany. Centralized via one "IT manager"
<DocScrutinizer05>
these are on top of our current geek customers
<DocScrutinizer05>
such thing could become amazingly huge
<DocScrutinizer05>
only downside: _we_ (as in: Neo900 group) can't finance nor drive such project. I can just offer support
dos11 is now known as dos1
<DocScrutinizer05>
and cooperation
<larsc>
security hardening is really hard to get right
<DocScrutinizer05>
sure, but we don't need to meet mil specs regarding that
<DocScrutinizer05>
;-)
<DocScrutinizer05>
just provide a somewhot useable and established UX aka "ecosystem" aka OS&app-collection that is already made with user privacy in mind
<larsc>
everything else is snakeoil
<DocScrutinizer05>
nah, our hw design already does 95% of the job for you
<DocScrutinizer05>
this is not meant to withstand a torture in NSA forensic labs
<DocScrutinizer05>
it just shouldn't give away all your contacts on a rogue SMS already
<DocScrutinizer05>
or keep them reasonably safe against the random spy with no special skills when device is lost
<DocScrutinizer05>
ensure that device is off when you switch it off, without need to remove battery
<DocScrutinizer05>
make sure that device doesn't eavesdrop on audio even when it is on
pcercuei has joined #qi-hardware
<DocScrutinizer05>
a LOT of people are interested in such a device
viric has quit [Ping timeout: 240 seconds]
<DocScrutinizer05>
add a cryptfs with password query on unlock/power-up to that, plus our awesome hw-driven tight modem monitoring
<DocScrutinizer05>
and you're basically done.
<DocScrutinizer05>
what you need to make sure: kick off all simple ways for user to install potentially rogue apps
<DocScrutinizer05>
simply done by setting up repos accordingly
<DocScrutinizer05>
and not allowing browser to install apps or even run JS or whatever. Best run browser in a sandbox/chroot
<DocScrutinizer05>
we don't need *high* security. We need reasonable setup of the wholoe system so users know it#s not THAT easy to catch a nasty virus or to spy on them without even doing a half day of preparing special tools
<DocScrutinizer05>
and you can post a looong list of things this device&OS will *not* do, unlike most other phones: call home, send your contacts list to anybody who asks for it, stay on while pretending it's off, allow your boss/fiance/competitor to simply install a 500EUR spyware during the 5 minutes you keft the device on table while going to poop, etc pp
<DocScrutinizer05>
s/keft/left/
<qi-bot>
DocScrutinizer05 meant: "and you can post a looong list of things this device&OS will *not* do, unlike most other phones: call home, send your contacts list to anybody who asks for it, stay on while pretending it's off, allow your boss/fiance/competitor to simply install a 500EUR spyware during the 5 minutes you left the device on table while going to poop, etc pp"
arielenter has joined #qi-hardware
<DocScrutinizer05>
believe me, customers are not interested in "100% security" - they know pretty well such thing doesn't exist, the least when it comes to mobile phones
<larsc>
first get things up and running then we can thing about the next step
<DocScrutinizer05>
and for sure they don't expect the device to be hardened against physical attacks in a lab, or similar threats that involve high effort
<dos1>
DocScrutinizer05: what about root access, usb, bootloader? those should be somehow hardened as well I suppose
<DocScrutinizer05>
sure, dos1
<dos1>
but yeah, with neo900 that shouldn't be hard
<DocScrutinizer05>
to a degree so that it's not simple to install spyware while you left the device to some rogue guy for a few hours, without you noticing later on
<dos1>
when thinking about "hardened fremantle" I initially still had n900 in mind with nolo etc. :D
<DocScrutinizer05>
a crypted rootfs already does a pretty good job for this
xiangfu has quit [Ping timeout: 252 seconds]
<DocScrutinizer05>
sure you probably *could* create a system that gets booted and runs your original system in a VM
<DocScrutinizer05>
so you possibly wouldn't notice the vast differences in behavior during boot (takes much longer, shows unusual displays for a while, etc)
<DocScrutinizer05>
but then such attack takes much more effort than installing a commercially available standard spyware
<DocScrutinizer05>
and had to get tailored to your particular device
<dos1>
just make a small system that mimics the passphrase dialog
<dos1>
should be managable in few hours if you really wanted to
<DocScrutinizer05>
for that you had to know what the passphrase dialog looks like, and you had to boot into the original system ag´fter querying the passphrase, in a way so user doesn't notice any differences
<DocScrutinizer05>
after+
<dos1>
had to know what the passphrase dialog looks like - easy, just boot it and watch :D
<DocScrutinizer05>
yeah, and then go back to lab and build it to your rogueware
<DocScrutinizer05>
and meanwhile user changed the way this dialog looks for him, maybe even automatically by a cronjob
<dos1>
you had to boot into the original system ag´fter querying the passphrase, in a way so user doesn't notice any differences - you just have to get the key once, so for some cases you may just force one unexpected reboot after that
<DocScrutinizer05>
again, how do you run such attack against a guy who gives you the phone for one hour?
<DocScrutinizer05>
or even for just 5 min
<dos1>
5 min - yeah, rather impossible
<dos1>
1 hour - might work
<dos1>
unless there are some additional things like cron-changed details you mentioned
<DocScrutinizer05>
I don't see that happen. No "applied forensics corp" will ever sell a "spy your fiance" app for 1500 bucks, that a noob can install on a Neo900 with "fremantle hardened edition"
<dos1>
yep
<DocScrutinizer05>
so, mission accomplished
<DocScrutinizer05>
again, nobody expects the thing to withstand a massive direct attack run by a group of experts with time and funds
<dos1>
but if your fiance is computer geek like us and he/she is extremely envious, it might work :P
<DocScrutinizer05>
people want reasonable protection against simple attacks your neighbour could run against you
<DocScrutinizer05>
when you think you can outsmart the CIA or NSA or whomever, when they got special interest in particularly you, then you lost the game before you started
<DocScrutinizer05>
and users pretty well are aware of that fact
<dos1>
sure, I just imagine that my neighbour may have for instance the same knowledge that I am
<dos1>
I have*
<dos1>
so if I can find some way to break into such device of my neighbour, then it's not good enough IMO :D
<DocScrutinizer05>
I promise you couldn't come up with a reasonable attack sttrategy that would allow you to install spyware on my Neo900 without me noticing
xiangfu has joined #qi-hardware
<dos1>
there's also the matter of what "without me noticing" means
<larsc>
is that a challenge?
<DocScrutinizer05>
that's pretty simple: any such attacker may not risk that *anybody* ever notices
<dos1>
for you some unexpected reboot while booting would be clear signal that something is wrong
<dos1>
DocScrutinizer05: and what if they are ready to risk? if ~90% possibility of not getting caught is good enough?
<dos1>
while for you reboot would be clear sign, someone else might not even think about it
<DocScrutinizer05>
create an attack you think might work, and we will discuss it. Until then I say what we need is a fremantle with cryptfs and browser in a snadbox, and a proper lockout of any way to install random apps that are not certified by the one who offers that hardening service
<dos1>
oh, another idea - instead of "unexpected reboot" just fake battery going flat
<DocScrutinizer05>
so what?
Codora has quit [Ping timeout: 260 seconds]
<dos1>
I mean, do that after typing the passphrase
<dos1>
then go into "act dead" mode while doing some heavy infinite loop on cpu to actually make the battery flat
<DocScrutinizer05>
again, then what?
<DocScrutinizer05>
then you maybe got the passphrase, and with much luck you actually were able to send it out via WLAN or WWAN
<DocScrutinizer05>
you still have not installed anything on the device
<dos1>
if you have passphrase, you have the rootfs
<dos1>
so together with sending passphrase you can setup remote shell as well
<DocScrutinizer05>
and you're still thinking of a simple comstant text passphrase
<DocScrutinizer05>
remote shell to YOUR system
<DocScrutinizer05>
not to the USER's system
<dos1>
to your system with rootfs of users system mounted and decrypted
<DocScrutinizer05>
haha
<DocScrutinizer05>
when you use some sort of challenge-response authentication, then you're pretty lost with your approach
<dos1>
yep
<dos1>
so it has to be taken into consideration as well :)
<DocScrutinizer05>
again, we are not interested in this sort of highly tailored-to-singular-target threat
<DocScrutinizer05>
when you think your authentication can get tampered by such attack, the the company who offers such service simply ships a new authentication every month
<DocScrutinizer05>
good luck with keeping your penetration tools up to date
<dos1>
if the scenario I described would be possible, then I'd consider it not hardened enough, as it's not very sophisticated attack IMO
<DocScrutinizer05>
to make it a tad more challenging for my fiance, we can roll out new authentication method/plugin in randomized timeframe, so nobody knows if the tool you want to use will still work or already make the whole rogue effort fire back at you
<dos1>
average joe couldn't do that on his wife's phone anyway, but average dos1 could :D
<dos1>
but if some challenge-response or whatever else makes it not possible, then it's fine
<dos1>
however, allowing to boot some different OS without authentication is still asking for a trouble in my eyes :P
<DocScrutinizer05>
nobody is going to boot your device while you're pooping, since they couldn't boot it to the original system to hide that they messed with device
<DocScrutinizer05>
and, thinking about it, we maybe could implement a boot counter so any booting of other OS would get detected by the still untampered original OS
<dos1>
yeah, just thought about it as well
<dos1>
and maybe some led blinking periodically that cannot be turned off programmatically to indicate that device is on?
<DocScrutinizer05>
that's already planned
<DocScrutinizer05>
though optionally
<dos1>
so without "fake off" possibility it should be much harder, yes
<dos1>
while allowing booting some other os might be handy in terms of recovery
<DocScrutinizer05>
fake-off is a pretty low concern to me, since average user will NOT turn off their Neo900 since it#s safe when on
<dos1>
DocScrutinizer05: but battery flat scenarios with fake-off might mask some unexpected behaviour
michael_lee has quit [Quit: Ex-Chat]
<DocScrutinizer05>
battery flat would be double-useless
<dos1>
I mean, faking battery going flat, so you're not suprised when it goes off
<DocScrutinizer05>
since other hw monitoring e.g. for modem-TX also kicks in when you pretend you're off
<dos1>
as opposed to sudden unexpected reboot which is causing you to be suspicious
<DocScrutinizer05>
so you pretend battery flat, while device LED signals not only device on but also signals constant TX activity
<dos1>
when LED signals device on then there's no point of doing battery flat faking
<DocScrutinizer05>
and a sudden battery-flat is as suspicious as is a sudden reboot
<dos1>
and one might come with some other things to do than modem tx
xiangfu has quit [Remote host closed the connection]
<dos1>
DocScrutinizer05: imagine that you went to party/shopping/wherever but forgot about your phone, allowing fiance to play with it
<DocScrutinizer05>
yeaah, and my fiance or neighbor might finally decide it's way easier to implant a hidden mike with own battery and all, rather than tamper my Neo900
<dos1>
hah :D
<DocScrutinizer05>
particularly since I immediately swap battery when it's flat
<DocScrutinizer05>
and immediately boot device up when it's down
<DocScrutinizer05>
which pretty much renders all your busy efforts void
<dos1>
other users might have other habits
<dos1>
so someone returns, "hey, why it's off", turns it on, loading, passphrase, loading, it fakes turning off from low battery, "ooh, battery went flat"
<DocScrutinizer05>
honestly, I ask you to implement such attack so it actually works, or it literally didn't happen
<dos1>
I'm not saying that it is or will be possible to implement such attack
<DocScrutinizer05>
users want a device that doesn't come with spyware preinstalled, no way to install spyware by surfing arbitrary websites, and not acting as expected to silent SMS and similar stuff
Coyo has joined #qi-hardware
<dos1>
I'm saying that it might be possible, so it should be considered
<dos1>
as it can be prevented by very basic things like some led indicators
<DocScrutinizer05>
and particularly no standard commercial tool to install that turns the device into a zombie controlled from remote
<DocScrutinizer05>
and they want their contacts and notes to be considerably safe when device is lost
<DocScrutinizer05>
and when it returns to them, they will consider it compromised anyway
<DocScrutinizer05>
when they leave the device 5min on table or one night in charger, they expect it to be up and running when they come back
<DocScrutinizer05>
and maybe we can train them to remove battery for 5 seconds before they power their device up after a few days it stayed off in a locker
<DocScrutinizer05>
this will all not help against a sophisitcated hichtech attack, but it will make your neighbor's life _much_ harder when he plans to mess with your Neo900
<DocScrutinizer05>
and for the customers I'm talking about, that's just a sufficient improvement of security over anything they could get now
Coyo has quit [Read error: Operation timed out]
<DocScrutinizer05>
swo: mission accomplished
<DocScrutinizer05>
so, even
<DocScrutinizer05>
security is not about 100%, it#s about teaching users where the limits of such security are
<dos1>
:nod:
jekhor_ has quit [Ping timeout: 246 seconds]
Coyo has joined #qi-hardware
Luke-Jr has quit [Read error: Connection reset by peer]
Luke-Jr has joined #qi-hardware
Luke-Jr has quit [Read error: Connection reset by peer]
<wpwrak_>
just try selling a good high-end B&W TV ... ;-)
<DocScrutinizer05>
meh, you don't want to say 512MB RAM is like B&W ?
<wpwrak_>
ok. color, but with VHF only :)
<DocScrutinizer05>
errr
<DocScrutinizer05>
as opposed to what? VHF + UHF?
<wpwrak_>
+ cable :)
<DocScrutinizer05>
so DVB
arielenter has quit [Ping timeout: 252 seconds]
<dos1>
let me check the amount of RAM on my Freerunner... :D
<dos1>
58 MB used, 64 MB free
<dos1>
could be more freed if opimd-notifier was in C instead of Python
<wpwrak_>
hehe :) so much about "512 MB is not enough !!!!"
<wpwrak_>
well, it's of course that you can never have too much memory
wej has quit [Ping timeout: 260 seconds]
wej has joined #qi-hardware
arielenter has joined #qi-hardware
heberth has joined #qi-hardware
heberth has quit [Ping timeout: 260 seconds]
heberth has joined #qi-hardware
<nicksydney>
DocScrutinizer05: Neo900 can be created as a dev board for students to learn about mobile phones
<nicksydney>
wpwrak_: how's the weather down there ?
<wpwrak_>
pretty good. warm, clear sky, a good summer day
<nicksydney>
DocScrutinizer05: if Neo900 were drawn out as a single sided design how big the board will be ?
<nicksydney>
wpwrak_: it's time to head to the beach and check out all those beautiful girls :)
<nicksydney>
wpwrak_: oh don't forget the vodka
<DocScrutinizer05>
nicksydney: infinite
<DocScrutinizer05>
you can mathematically prove that you need at least 6 layers to un-nest a BGA of SoC's complexity
heberth has quit [Quit: leaving]
<wpwrak_>
the beaches are way too crowded. there are better places to see girls :)
<wpwrak_>
(6 layers) where ought to be a lot of other parameters if this is supposed to be true :)
<wpwrak_>
s/where/there/
<qi-bot>
wpwrak_ meant: "(6 layers) there ought to be a lot of other parameters if this is supposed to be true :)"
<DocScrutinizer05>
I'm desperate from openmoko.org MX broken with my SMTP mail sending
<DocScrutinizer05>
well, the other parameter is ball pitch, and obviously footprint at large
<DocScrutinizer05>
it's simply ridiculous to try and route a fine pitch ball grid array with some 6 or more concentric rings of balls on a single or double layer PCB
<DocScrutinizer05>
there's a certain minimum width of traces and a minimum diameter for laser vias
<wpwrak_>
well, single layer sounds more like a form of art if not magic ;)
<DocScrutinizer05>
(for double layer obviously, for single you don't have vias)
<wpwrak_>
the limiting factor may be power. even if you manage to get everything outside with, say, two layers, you'd still have very complicated paths and very long power routes. that can't be good for high-speed circuits.
<DocScrutinizer05>
so you can see easily how you need to route some 5 or 6 traces between two balls of the 0.4mm(?) FPBGA
<wpwrak_>
well, unless what have is a huge fpga of which you're only using a small number of i/os. then you may be able to get away with it.
<wpwrak_>
0.4 mm sounds suckish :)
<DocScrutinizer05>
not even then, since you can't place balls on solder stop with traces underneath instead of pads
<wpwrak_>
yeah, you need a pitch that lets you at least route one trace between balls
<DocScrutinizer05>
I think TI itself says for OMAP4 you need 8-layer at least, to un-nest
<wpwrak_>
how much easier life could be if someone made and sold decent pops :)
<DocScrutinizer05>
for me right now it would make me perfectly happy when I could find out what the heck is breaking my email
<DocScrutinizer05>
yesterdsay it temporarily started to work absolutely flawless again, then eventually broke and now it#s back to fubar it been yesterday morning
<DocScrutinizer05>
I tested with thunderbird as well, and same result
<DocScrutinizer05>
[2014-01-08 21:41:22] <pabs3> DocScrutinizer05: hmm, succeeded in sending myself a mail using swaks and a new test account: swaks --tls --auth --to pabs3@bonedaddy.net --server smtp.openmoko.org
<DocScrutinizer05>
I'm totally puzzled even which port to use
<DocScrutinizer05>
while it seems like smtp is port 25, thunderbird started stating something like 564 or whatever for the default of STARTLS
<DocScrutinizer05>
pabs3's line above doesn't really state which port and which exact authentication etc he used
larsc has quit [Ping timeout: 252 seconds]
nicksydney has quit [Read error: Connection reset by peer]
lekernel has quit [Quit: Leaving]
nicksydney has joined #qi-hardware
rz2k has joined #qi-hardware
<nicksydney>
i'm sure someone here can help me to understand something inside Kernel :)
<nicksydney>
most of the time i read about graphics driver and it says that the vendor closed source it....so was wondering when the vendor closed source the driver does this mean that the kernel only knows how to load the driver (.so) file and then use like a virtual function to call the functions inside the closed code ?
<wpwrak_>
typically the kernel code of closed drivers is relatively simple and open. but the code that does the real work runs in user space, talks to that kernel driver, and is closed.