<wpwrak>
aw: the solder mask clearance (green) clearance around the pad (purple) is 5 mil. the solder mask openings of the two components almost touch. is this likely to be a problem ?
<wpwrak>
aw: if yes, should i reduce the clearance to, say, 4 mil ?
<aw>
the red square is your solder pad? the purple is your copper? the black is your solder mask?
<wpwrak>
black is pcb. purple is copper. pink is solder paste. green is solder mask.
<wpwrak>
the dark green (only on the first png) is the bottom layer
<aw>
okay , so now your clearance setting is 5 mil between solder mask and copper (purple) is 5 mil. right? so 5 + 5 = 10 mil which is your clearance between solder mask and solder paste!
<aw>
right? i see like that.
<aw>
correct me, if wrong!
<wpwrak>
solder paste has only 1 mil from copper. if i increase the solder paste clearance, the solder paste would basically vanish.
<aw>
um...i see know
<aw>
first a very 'safe' way is : solder paste can have 5 mil fro copper, then copper is 5 mil from solder mask. this is the most safe manufacturing.
<aw>
but in case,
<aw>
you actaully no need to care that 1 mil at all, just make a clearance 5~ 10 mil between solder paste and solder mask.
<wpwrak>
hmm, solder paste 5 mil from copper looks like trouble with qfn. there, there the pads are only 9.5 mil wide ...
<kyak>
< wpwrak> & asked the sailor upon entering madam's parlour :) <-- what happened next, can easily be found on the Internet :)
<aw>
now your two side by side yes is two closer.
<wpwrak>
5 mil is what i have now. so far, so good :)
<wpwrak>
aw: you mean they're too close ?
<aw>
i see your both green is 2 mil clearance (both solder mask). It's very bad!
<wpwrak>
aw: it's the pair below the balun (on the atben boards)
<aw>
wpwrak, just make both solder masks is 5 mil, is it necessary to calibrate the performance after shift 3 mil?
<wpwrak>
kyak: yeah, these days, the sailors don't go there anymore. they just connect their laptop ;-)
<wpwrak>
aw: performance should be the same. not sure if i have enough room, though ...
<aw>
wpwrak, "5 mil is what i have now. so far, so good"Â Â >>> which is you said?
<wpwrak>
aw: that was about the pad to solder mask clearance
<aw>
from your picture I can see the clearance from both solder masks.
<aw>
is 2 mil only.
<wpwrak>
yes
<wpwrak>
i wonder if i should simply lower the solder mask clearance from 5 mil to 4 mil. that would still be slightly more than the 0.1 mm usually assumed for layer registration accuracy
<wpwrak>
and it would increase the gap to 4 mil as well -> problem probably solved ;-)
<aw>
yup, make it added to 5 mil please. No easy that makers can easily control it as all pcbs as same good enough as 2 mil AFAIK.
<aw>
no.
<aw>
as your pcbs is only two layers. I'd like yours to be more producible in everywhere. so if you always keep 5 though, then you have no troubles on this more!
<wpwrak>
except squeezing in the components ;-) let's see ...
<wpwrak>
phew. i think it just fits.
<roh>
hm. has anybody seen john?
<roh>
from H:D
<aw>
wpwrak, surely if you lower solder mask to 4 mil, finally your clearance between solder masks is 4 mil though.
<aw>
from my pcb makers they told me before. As a nominal /central value they like 5 mil, if you make design as 4 mil, an 'aggressive' pcb sales who want to deal with you , they will say no problem. but actually if their pcb capability is NOLY said as 3 mil, then don't trust them at all.
<wpwrak>
aw: now i have about 6 mil.
<aw>
because of a known etching tolerance.
<aw>
wpwrak, hmm...great.
<wpwrak>
aw: (4 mil) pretty soon, the components will be smaller than their tolerance then ;-)
<wpwrak>
it's a bit old but gives you and idea of the basic concept
<aw>
wpwrak, not yet..seeing..okay..good
<qi-bot>
[commit] werner: Added a new pad type: trace pads, for antennas and other trace-like elements. http://qi-hw.com/p/fped/49a1bbd
<qi-bot>
[commit] werner: - gui.c (change_world_reselect): take into account that we may get called http://qi-hw.com/p/fped/1409cfa
<xiangfu>
aw: checking... and updating  :)
<wpwrak>
ah, catching up ;-)
<aw>
wpwrak, so how do  *.deb files under downloads folder make sure it's newest to KiCad sources? or it's made manually. AFAIK wolfgang made that for me.
<wpwrak>
aw: fped isn't connected to the kicad sources. as long as kicad doesn't change the board file format, fped should be compatible
<xiangfu>
aw: I am update the fped  .deb file now.
<aw>
xiangfu, wow...it was you?  great.
<xiangfu>
aw: hmm... but I only make the fped deb file  not kicad
<aw>
xiangfu, ok, tks.
<xiangfu>
seems there are not so many people in this IRC lives in GMT+8Â Â :D
<qi-bot>
[commit] werner: Added a new pad type: trace pads, for antennas and other trace-like elements. http://qi-hw.com/p/fped/49a1bbd
<qi-bot>
[commit] werner: - gui.c (change_world_reselect): take into account that we may get called http://qi-hw.com/p/fped/1409cfa
<wpwrak>
the only thing missing are the drill holes
<kyak>
xiangfu: i'm close to you, this week the time shifts to GMT+4 in Russia and stays liek this forever (i.e. no more daylight saving time in Russia)
<wpwrak>
kyak: celebration for a week ?
<kyak>
of course, it's not GMT+4 throughout the whole Russia :) we've got 6 or 7 time zones, if i'm not mistaken
<kyak>
wpwrak: we don't need an ocassion to celebrate ;)
<tuxbrain>
i'm running  yes "" | make oldconfig now and will retry
<panda|x201>
wolfspraul, ping
<xiangfu>
tuxbrain: thanks. I am not test those steps. just write them by memory. sorry. fixed now.
<xiangfu>
tuxbrain: the 'yes "" | make oldconfig' have to run twice. first before and after 'make package/symlinks'
<xiangfu>
tuxbrain: sorry. it's have to run again after 'make package/symlinks'
<xiangfu>
no need in before.
<tuxbrain>
xiangfu: no need to sorry dude , that's, us, the testers are standing for, isn't it? :)
<xiangfu>
tuxbrain: I am updating the wiki page, maybe this is clear : [ first time we download this '''.config''' file is for 'package/symlinks' don't bring up the 'make menuconfig']
<xiangfu>
:D
<tuxbrain>
ok let me know when you finish (please also the release part) and I will review/modifiy in needed
<tuxbrain>
mmm running doen't make any difference
<tuxbrain>
running 'yes "" | make oldconfig
<tuxbrain>
should I rerun 'yes "" | make oldconfig'?
<xiangfu>
1. downlad the feeds.conf  2. run ./scripts/feeds update -a ./scripts/feeds install -a 3. download the .config again. 4. run [yes "" | make oldconfig] 5. then if you open .config you can see a lot of packages have selected.
<xiangfu>
tuxbrain: wait. and also ./scriptfs/feeds install -a
<xiangfu>
the 'make package/symlinks' = ./scripts/feeds update -a && ./scriptfs/feeds install -a
<xiangfu>
the only different between them is 'make package/symlinks' will check if you have a ".config" if there is no ".config" file it will bring you the 'make menuconfig'
<tuxbrain>
good news now seems to work, bad news supertux build fails
<tuxbrain>
Checking 'pngcrush'... failed.
<tuxbrain>
supertux: Please install pngcrush.
<xiangfu>
you need install pngcrush in your host system
<tuxbrain>
another thing to modify in the wiki :) need to be added to the apt-get install
<wpwrak>
tuxbrain: btw, did you get a quote from the smt fab yet ? and did they say whether they need a stencil or not ?
<tuxbrain>
wpwrak: beautifull pics but I don't understand a thing (well a lot of but I will ask just one)  what is ligth green and red? 2 sides?  isn't the circuit one side?
<tuxbrain>
I have ping them last night , due I have not recieved any answer yet
<tuxbrain>
if I will not recieve any answer in a couple of hours I will phone them
<wpwrak>
tuxbrain: (light green/red) that's front and back, to show the connectivity. both sides of the board are used for traces. components are only on the front.
<kyak>
xiangfu: so actually you marked "useless-in-trunk" those patches that went into data/* ?
<xiangfu>
kyak: yes
<kyak>
xiangfu: it's a very good idea, now there is only alimited number of patches that can be worked through one by one
<xiangfu>
kyak: yes.
<tuxbrain>
wpwrak: ok,  beautifull images  in deed
<kristianpaul>
wolfspraul: (quick update) I'm moving about porting the namuru vreilog core to the mm1, if it works i could really _speed_ up some TODO about gps-sdr
<kyak>
then you should be able to reproduce the release build. The only problem is that at the time of release build, this problem wasn't fixed yet.. I had a temporary patch for that that you could use
<kyak>
(xiagnfu didn't have such problem because his version of autotools match)
<kyak>
tuxbrain: you might have to modify digits to match exactly your version of autocrap
<kristianpaul>
okay i gotta go office. read you later
<tuxbrain>
kyak, a more detailed instructions on how to apply the patch will be much apreciated...
<kyak>
cd feeds/packages
<kyak>
patch -p0 < ../../pango.patch
<kyak>
giving that you saved the patch in pango.patch of your openwrt-xburst directory
<kyak>
*top directory
<tuxbrain>
ok
<kyak>
you can modify the feeds/packages/Xorg/lib/pango/Makefile manually as well, it would be faster in this case than saving/applying the patch :)
<tuxbrain>
mmm patch: **** Only garbage was found in the patch input.
<tuxbrain>
well I think I will do manually
<kyak>
yea
<tuxbrain>
but this sould be fixed someway, people must be able to build the releases without such kind of errors....
<kyak>
build the latest, it should be fixed
<kyak>
you will face other problems though
<kyak>
but then there are bug reports for those already
<tuxbrain>
kyak not good policy , release should mean, "hey! here following that steps you are save", I agree that if you play with the edge lastets development you should find issues, but a clean build should be must for releases
<kyak>
there is always some number of people/configurations where the build will fail
<kyak>
if not for this reason ,then for something else
<kyak>
we would have to test that release build fine on all major distros and on x86/x64
<kyak>
it's not possible
<kyak>
og yeah, i remember this shit
<kyak>
it tries to links against host libgobject
<tuxbrain>
kyak: but the errors I'm founding are no matter of arch as you can see so your reply doen't convince me, don't get me wrong, I'm really thakful for the work you , xiangfu , Jay7 and whatever contributor does , I'm just putting on the situation of the new commers or , to some one that whats to use actual releases as base to develop a dedicated product... with such fails they wll just think, bah doesn't work and go for other solution, we must avoid thi
<tuxbrain>
s on releases.
<kyak>
but it builds find on buildhost
<tuxbrain>
and what makes build host different ? why is not on the requisits on the wiki, what step I forgot to follow?
<kyak>
i remember problems that are arch specific.. like gcc-mips problems. i remember problems that are distro spoecific, like used version of host autotools. There can be even such problems for two people on the same arch/distro when one of them has build some package (like perl) before, and then another package (like plplot) would fail :)
<tuxbrain>
any solution for the libgobject ? :(
<kyak>
trying to remember &
<tuxbrain>
propose a clean test from scratch in the build host of a release before relesase it.
<tuxbrain>
clean test build
<kyak>
this is done already
<kyak>
the clean build
<kyak>
why don't you believe that such problems prove to be so very specific?
<kyak>
tuxbrain: i can't really find or remember the patch for that libgobject problem.. maybe we could try to dig a little bit..
<tuxbrain>
kyak: I'm reluctant to believe than two x86 machines sourcing form same repos, starting in same point , with same instructions give diferent results... one of the suposits must be false I want to to figure out what is the false and solve it
<wpwrak>
tuxbrain: are you developing a prototype or do you already have something finished ? in the former case, maybe check if the things you need are already available on jlime, and if so, do the development there
<wpwrak>
tuxbrain: you can always brave the openwrt build process later (or, better, outsource the problem ;-)
<tuxbrain>
wpwrak: well my intentions where have a stable opentwrt build based on the current release for multiple reasons:
<tuxbrain>
have a toolchain with libs avialable courrently on the release.
<tuxbrain>
with that toolchain set up Qt-Creator to build/upload/run for/on NN
<tuxbrain>
helping in create proper packages for the avr tool chain
<tuxbrain>
create a proper package for Nanokoder a KISS(due I don't know more) code editor based on QT with Nanonote constrains in mind, line numbering , code Hightligh, zoomamble font and  able to run external scripts
<tuxbrain>
meanwhile the first one is not posible I'm working in the last one
<wpwrak>
tuxbrain: yeah, i meant the work on nanokoder. maybe worth checking if jlime has all the libs you need already. once you have something that works, the openwrt gang should also be able to help you more efficiently
<tuxbrain>
line wraper on/off, line number, font zoom already working , no I will go for code highlight(C++ for now), then open/save files, then execute scripts, then reuse of code highlight definitons of nano or joe, then jump to line from compile error log, then support custom profiles, then I will be rich and will make others to do what I want
<tuxbrain>
no menus, no dialogs, just key press, and a text widget...
<wpwrak>
highlight should be fun :)
<wpwrak>
(no menus, etc.) the return of king vi to arduinoland ? :)
<qi-bot>
[commit] Werner Almesberger: prettygerbv: restructured project file generation to skip missing layers http://qi-hw.com/p/eda-tools/4e5278a
<wpwrak>
kristianpaul: viric wrote "that should make [...] 'c' how many values" ;-)
<kristianpaul>
ahh
<wpwrak>
google rocks for figuring out unruly constructs :)
<kristianpaul>
ok ok!!
<kristianpaul>
:-)
<wpwrak>
well, google plus the sites with the actual answers ;-)
<kristianpaul>
;-)
<kristianpaul>
wpwrak: how do you worked when no google existed?
<wpwrak>
kristianpaul: slowly :)
<kristianpaul>
hahah
<kristianpaul>
searching at stackoveflow
<viric>
wpwrak: no no
<viric>
c will be *how many values have been read*
<wpwrak>
viric: yes, that's how i understood it. basically like fread in C.
<viric>
[v,c] = ...  is the octave way of getting the *two results* a function may return.
<viric>
v = f()Â Â Â Â f returns one result
<wpwrak>
viric: (C) only that you get a variable and don't pass a buffer
<viric>
[a,b] = f()Â Â Â Â f returns two results
<wpwrak>
well, s/variable/value/, to be precise
<viric>
wpwrak: well, octave does not give memory pointers to users that easily :)
<wpwrak>
viric: real men hard-code their pointer addresses ;-)Â Â Â Â int *buf = (int *) 0x810ad4f60;Â Â /* buf = */ malloc(1234);
<kristianpaul>
:-|
<viric>
wpwrak: that's a weird 64-bit pointer :)
<wpwrak>
viric: yeah, it admittedly is :)
<wpwrak>
viric: corollary: real men also hack their libc to produce just this sort of pointers :)
<viric>
I'm waiting for an amd32 abi...
<viric>
x86_64 instructions and registers, all with 32-bit pointers
<viric>
I tried, and firefox takes 50% less RAM in 32-bit (in a single experiment I did)
<wpwrak>
wow :)  do they have something like  typedef long number;  ? :)
<viric>
hm no :)
<viric>
why would they?
<wpwrak>
well, that would explain why going from 32 to 64 bits doubles the size of everything
<viric>
ah
<viric>
only long long is 64-bit in linux
<viric>
in gcc, basically
<wpwrak>
that is, unless they mainly store pointers ...
<viric>
And the bigger use of memory comes through bigger pointers, not bigger integers
<wpwrak>
long long is 64 on 32 and 64. long is 32 on 32, 64 on 64.
<viric>
It's quite usual for many programs to store lots of pointers
<wpwrak>
makes one wonder what all these points point to. more pointers than actual data. does that sound like excessive bureaucracy ? ;)
<kristianpaul>
lol
<viric>
It's a usual phenomena when you use more memory to gain speed :)
<wpwrak>
more administration -> higher speed. qed. ;-)
<viric>
You keep pointers to places you want to access often, instead of looking for them every time
<tuxbrain>
real man do their programs with wire and void valves, not that fancy and fragile perforated cards...
<viric>
sewing memories
<tuxbrain>
was abducted by the infinite classes structure of Qt, his headadache++ reading sentecences like "long long is 64 on 32 and 64. long is 32 on 32, 64 on 64."
<Jay7>
use python :)
<tuxbrain>
I will do so extensively on Ya, on ben python+qt+building in back groundgroud would be to much, also graphical libs and python is also a good mambo jambo to end with a great headache
<kristianpaul>
:-)
<kristianpaul>
real men should not be talking here about real men :-)
<kristianpaul>
hides
<tuxbrain>
real man do not hide from real men
<tuxbrain>
thinks we need more woman on qi-hardware :)
<kristianpaul>
wpwrak: in atbb/atusb how do you avoid the atmega/benanote inject noise to the RF chip?
<kristianpaul>
wpwrak: Do you have a particular manipulation process with the RF chip itself before and during rework?
<wpwrak>
kristianpaul: there are no special precautions for noise filtering
<wpwrak>
kristianpaul: (manipulation) err, what kind of manipulation did you have in mind ?
<kristianpaul>
wpwrak: to avoid ESR?
<kristianpaul>
wait no
<kristianpaul>
electro static discharges
<wpwrak>
kristianpaul: ESD ... during assembly ? naw. buenos aires has a pretty humid climate all year long. there's not much of a risk of my frying a chip with ESD.
<kristianpaul>
ok
<kristianpaul>
hmm
<wpwrak>
kristianpaul: i have ESD protection in atusb on the USB pins. nothing on atben, though. we'll see how this goes :)
<wpwrak>
tuxbrain: if you prefer just formatted text (like the "shopping list"), i can generate this as well
<kristianpaul>
wpwrak: i hope next month i'll have some money for a 4 pcb run for atben (just for fun and learn)
<tuxbrain>
wpwrak: as I said I already done the matching and already send the list to smt vendor, but sure this one sould be better to pass as production files :)
<wpwrak>
tuxbrain: oh, and what sort order do you prefer ? component reference ? by quantity ? by digi-key part number ? don't care ?
<tuxbrain>
wpwrak: don't care I think
<wpwrak>
tuxbrain: okay. i'll sort by component reference then, like we did in openmoko.
<tuxbrain>
kristianpaul: common, I hope in a month I will be plenty of atbens for you
<tuxbrain>
wpwrak: ok
<kristianpaul>
tuxbrain: ;-)
<tuxbrain>
save the money to come to spain and catch them your self :)
<kristianpaul>
hehe :-)
<kristianpaul>
That will be nice, but VISA is hard for most Colombian's espacially for spain, you may already know mostly why ;-)
<kristianpaul>
btw !
<kristianpaul>
well may be not, but if some body can cath a zarlink gp2021 why can bring some gps stuff for the nanonote too in a hackish way
<kristianpaul>
of course this is off-topic for what i'm currently working with SiGE Front End
<kristianpaul>
but this was a gps hack toy in the 90's
<wpwrak>
there are real maps that are drawn in this style
<wpwrak>
tuxbrain: files updated. now sorted by component reference
<kristianpaul>
sure, but that size and pocket thing??
<kristianpaul>
loves linux networking versatilly and chroot
<tuxbrain>
too
<tuxbrain>
btw the "ye-olde-gps" was from 1920 :)
<kristianpaul>
nice
<wpwrak>
yes ! commit 6000 at svn.openmoko.org ! ;-)
<kristianpaul>
you?
<kristianpaul>
who else ;-)
<tuxbrain>
what is there?
<kristianpaul>
tuxbrain: wpwrak repo !!
<kristianpaul>
very valuable
<wpwrak>
tuxbrain: the core parts of the BOM processor (trunk/eda/boom/)
<kristianpaul>
make a copy of it !
<tuxbrain>
yep, please backup whatever you have under openmoko domains
<wpwrak>
kristianpaul: (copy) yeah, when openmoko.org dies, i'll migrate boom and fped over to qi-hw
<wpwrak>
tuxbrain: i have local copies of everything :)
<tuxbrain>
projects seem totally down, isn't it?
<kristianpaul>
wpwrak: and the other stuff? i saw a memoryspeed tester too
<wpwrak>
tuxbrain: (projects) dunno. it hasn't been well for a very very long time already
<wpwrak>
kristianpaul: i have a copy of the entire repository. but fped and boom are probably the most valuable parts right now
<kristianpaul>
sure
<wpwrak>
(oshw logo) the "Iconographic Microchip" looks more like "broken" than "open" ;-)
<wpwrak>
and so do all of the "Geared" thingies. remind me of drawings of body cells bursting after some virus infection
<tuxbrain>
wolfspraul: definitively we need more visibility (qi-hardware) I have read an open hardware article on farnell magazine, and apears Arduino and ... BugLabs!!!
<tuxbrain>
of course mention to the broken chip icon
<tuxbrain>
sorry oshw
<wpwrak>
hmm, ordering by component reference isn't quite clean. C10 comes before C2. not good. let's fix this ...
<wpwrak>
tuxbrain: qi-hw ... isn't that this company that makes some sort of clamshell display with keyboard for the arduino ? :)
<tuxbrain>
wpwrak: nah, that was linuxbrain or somthing like that qi.hw was the one of the milk dispenser under fpga
<kristianpaul>
qi-pixel?
<wpwrak>
kristianpaul: is that a zombie penguin ?
<tuxbrain>
wow a tux-borg
<wpwrak>
i like the oshw "geared" logo. it's exactly the style in which biological cells are drawn. in the first one, we see a cell, complete with nucleus, where the membrane has been damaged. below, we see more damaged cells, each with a hydroxyl radical inside, which is quite fitting, see also the last paragraph of http://en.wikipedia.org/wiki/Hydroxyl#Hydroxyl_radical
<tuxbrain>
I like the copyleft one, but then our motto will be really attached to openhardware(tm)
<wpwrak>
yeah. let's vote against it :)
<tuxbrain>
yay code highlighter working :)
<wpwrak>
tuxbrain: /* " */ "does this work ?"
<tuxbrain>
yep :)
<wpwrak>
tuxbrain: //* and how about this ?
<tuxbrain>
mmmm //* is recognized as /*
<wpwrak>
(-:C
<tuxbrain>
bastard
<wpwrak>
that was easy ;-)
<tuxbrain>
damn it qt-creator does it well...
<tuxbrain>
ok //* fix for next release :P
<tuxbrain>
let's go to open/save files
<tuxbrain>
also for next release limits on zoom in zoom out, I can zoom out to 1 point font and to max_int :P
<wpwrak>
kristianpaul: ah, another thing worth saving from openmoko.org is my TMC stuff (lab instrument control)
<tuxbrain>
any ones knows if QFileDialog will work on NN and how it should look like?
<tuxbrain>
on pc it opens the system file dialog, so I'm afraid in NN will not work...
<qi-bot>
[commit] Niels: Revert "first try to fix building with OpenWRT toolchain that does not include QtCore and QtGui in the include path" http://qi-hw.com/p/nanomap/f1a2430