<wolfspraul>
wpwrak: my report about case insensitive matching in the field contents or part number matching was wrong. no need to investigate. sorry I was confused about a few things, but it's all clear now and all those comparisons are case sensitive.
<wolfspraul>
wpwrak: in boom/README, line 145, I think the two columns are flipped. The KiCad field name is 'Footprint' and the boom field name is 'FP'.
<viric>
wpwrak: yes, I think so. Total success
<viric>
wpwrak: if I could read the program in the f321, would that be through usb, or there are some serial lines to do that?
<viric>
wpwrak: my previous attepmts at rev-engineering the usb protocol were troubles with linux < 2.6.21, that did not have a system (without patching) that captured *whole* frames. in the 2.6.35 I use I could do that. And luckily wireshark handles very well USB, and I did not have to solve the usb frames directly from usbmon.
<viric>
Did ever someone here play with the con tiki OS?
<viric>
and, for the nanonote, is there anywhere the distribution of the cost on the device parts?
<viric>
so, something like 30% the plastic case, 15% the microcontroller, 10% the screen, ...
<wolfspraul>
viric: very hard to say because you cannot break these numbers down to 1 unit
<wolfspraul>
the tooling costs way over 50,000 USD
<wolfspraul>
a very rough BOM I would say maybe 40 USD total
<viric>
uhm I don't understand those terms. 'tooling'? BOM?
<wolfspraul>
if you want to make 10,000, you can probably get it done for 400,000 USD
<wolfspraul>
oh sorry
<wolfspraul>
then even more so, explain to me what you want to do?
<wolfspraul>
if you want to make 1 NanoNote yourself, figure thousands of USD :-)
<viric>
no no.
<viric>
It was just curiosity
<wolfspraul>
if you want to make 10,000, like I said maybe 40 USD / unit, well, for the first 10K maybe 50 USD / unit with all the things that will go wrong.
<viric>
I realized that I have no idea where its price comes from.
<wolfspraul>
well the more you dig in the more the prices 'disappear'
<wolfspraul>
but that's because you are skipping over companies that you will need if you really want to do it
<wolfspraul>
the CPU costs maybe 3 USD
<viric>
but there has to be someone having done a process that leads towards the current price
<wolfspraul>
yes sure
<viric>
Is that written somewhere?
<wolfspraul>
no, just gut feeling and I believe we can get things started this way
<viric>
So, I know a bit electronics, a bit programming, assembling, ... but I know very little on the 'production' side.
<viric>
I thought that if I could see an open process about that, I'd learn something.
<wolfspraul>
ah yes
<wolfspraul>
then you are right here :-)
<viric>
:)
<wolfspraul>
so there are multiple moving pieces
<wolfspraul>
we started with the Ben NanoNote where a number of normal (closed) companies are helping us
<wolfspraul>
that's so that we have some 'gravity' that pulls the project around real user needs
<wolfspraul>
now we are using both that user gravity, as well as our copyleft goals, to move things forward
<wolfspraul>
so we are building out the server infrastructure, we started to co-work with Milkymist, there is hacking going on around the case, GPS, RF, etc.
<viric>
aha
<wolfspraul>
well it's a lot, and these things all belong/fit together but admittedly a bit hard to see
<viric>
I imagine.
<viric>
So you don't have much meetings with slides, pie charts, and all that? :D
<wolfspraul>
so unless you are very patient (which the people buying a NanoNote or even in this channel mostly are), I think it will only become more succesful if we can make some really cool/different products.
<wolfspraul>
ah no [slides]
<wolfspraul>
scratch that
<wolfspraul>
:-)
<viric>
hehe
<viric>
what on the investors?
<wolfspraul>
no investors, just savings and time donations
<viric>
great.
<wolfspraul>
the best advise I got on this project was "earn your way"
<wolfspraul>
advice
<wolfspraul>
so I follow that advice
<viric>
ok
<viric>
I may some day engage in all this.
<wolfspraul>
wonderful
<viric>
What I do now I consider "not engaged in it" :)
<viric>
wolfspraul: do you know 'fossil'? If you know, you will understand the whole rss system at once. If not, then it will be harder.
<wolfspraul>
no I don't know
<viric>
well, it's a distributed version control system, where some actions you do with command line (as svn, git, ...), and for some others it has a small web server, and then you do the tasks with a web browser.  It has not only distributed version control of code, but also wiki, ticket system, ...    Then I wrote a php script that fetches feeds and writes them in 'wiki files', than then the usual fossil webserver will serve nice for lynx to browse.
<viric>
10 lines. No magic. Nothing that even can be integrated in any distribution, I guess. :)
<viric>
wolfspraul: so it's more about putting things together and using them together, than having a program named 'distributed RSS reader'.
<viric>
wolfspraul: as you see, I have that in a web server. So I can read the same from a computer with internet access (browsing to my server page), with my PC (that has a copy of the repository synchronizable), or with the nanonote, that can be synchronized with the PC.  Three places of reading.
<viric>
I imagine everyone does whatever he likes to, with the nanonote :) wpwrak makes boards and boards for it. hehe
<viric>
wpwrak: using the sdio as a serial line with gpio userland controlled, what speeds did you achieve?
<B_Lizzard>
With sound enabled, shouldn't the /dev/snd devices be populated via udev?
<viric>
B_Lizzard: I think so.
<B_Lizzard>
Ugh, they're not in 2.6.32 and not in 2.6.36
<B_Lizzard>
Hmmm, I'm using udev 124
<B_Lizzard>
Latest is udev 164
<B_Lizzard>
Lemme try with a newer udev version
<wpwrak>
wolfspraul: (case sensitive match works as intended) thanks !
<wpwrak>
wolfspraul: (FP) fixed. thanks !
<wpwrak>
viric: (usb analysis) maybe you could wire a post about this. the process could be interesting for others too
<wpwrak>
viric: (sdio speed) i didn't measure the peak thoughtput. i tested a 1 pin toggle and got about 1 MHz. so expect something like 300-500 kHz for SPI-like one-way data. maybe 200-300 kHz for SPI-like bidirectional
<dandon>
any laptop users? would go with Dell Inspiron 15R?
<logictheo>
I have a netbook.
<viric>
wpwrak: It looks like my free dyndns domain expired *TODAY*. I just payed for renewal.
<viric>
wpwrak: I had it almost 8 years....
<viric>
wpwrak: it should work now.
<viric>
wpwrak: you were the first to notice the domain disappeared.
<wpwrak>
nice ! maybe link to a place that explains how to watch usb with wireshark (if there's one) ? also, what doesn't work with VIA ? the usb device or just the monitoring in qemu ?
<wpwrak>
(first to notice) yeah, the bugs flock to me ;-)
<viric>
wpwrak: I think I'll put you in the 'thanks' list, helping me with the chips. What is your naem?
<viric>
name
<wpwrak>
i didn't do much :) my name is Werner Almesberger
<viric>
wpwrak: I wrote to the manufacturers of the PCE device, so they may want to mention that "it may work on Linux too with this".
<viric>
I have a coworker named Werner
<wpwrak>
(wireshark) nice ! didn't know it was that easy
<wpwrak>
(manuf) hehe, "and, by the way, your secret protocol is no secret anymore" ;-)
<wpwrak>
ah, it's not a very common name
<viric>
ah, I don't think they will care. It's really really simple, and there is little magic on the PC software. All the hard parts are on the mcu.
<viric>
wpwrak: are you rumanian/hungarian/german at once, like my coworker? :)
<wpwrak>
hmm .. now an electrical engineering puzzle: i have a 50 Ohm transmission line, about 1-2 m long. my termination is something like 1 MOhm, and i can't just toss in a resistive divider because i need to (more or less) preserve the amplitude. the signal is aperiodic. if there *anything* i can do ? :)
<viric>
you can't change the termination?
<viric>
well, you want to adapt it to the line I guess.
<wpwrak>
viric: uh, at least two generations deep austrian, i think. then it gets a bit scattered. austria, germany, czech republic (but that was austria too, back then)
<wpwrak>
how do i change the termination without loss of amplitude or messing up my aperiodic signal ? :)
<viric>
let me take my notes on transmission lines
<wpwrak>
if i add a 50 R to GND, the signal is beautiful, but too weak
<wpwrak>
if i add an impedance matching network, the signal is beautiful too, but only if periodic
<wpwrak>
(well, beautiful in qucs. you never know what may happen in real life ;-)
<viric>
what is qucs?
<wpwrak>
a cicuit simulator. think spice minus the painkillers
<viric>
:)
<wpwrak>
well, or spice but after the invention of fire, wheel ;-)
<wpwrak>
qucs is great. its schematics editor is even easier to use than kicad :)
<viric>
I'm going back to what I once knew about lines...
<viric>
the impedance of the line is mostly capacitive
<viric>
around the frequency of your signal
<viric>
right?
<viric>
uhm, both capacitive and inductive.
<viric>
I think the 50Ohm are defined at a specificed signal frequency.
<viric>
How do you write the transmission line in that 'qucs'?
<wpwrak>
i put a transmission line symbol, with Z=50 Ohm and L=2000 mm
<dandon>
hm i didn't really look xD ^^. Dell Inspiron 15R classifies as notebook?
<dandon>
*netbook
<wpwrak>
not sure about frequency dependency of the characteristics imedance
<dandon>
dell.com
<wpwrak>
i thought it wasn't frequency dependent (of course, all the rest usually is)
<viric>
wpwrak: so, you need to know a model of a transmission line, and use capacitors and inductances to adapt it
<wpwrak>
i tried that - with a quite horribly distorted signal
<viric>
wpwrak: there is also about 'adapting lines'
<viric>
page 23 has a problem about it
<viric>
explained in page 21
<viric>
so you need a 'coefficient of transmission' of 1:Â Â tau = Gv(1 + rho)
<wpwrak>
i guess now i'm really confused ;-)
<viric>
hehe. Well, it does not look easy :)
<viric>
wpwrak: for what I understood so far, it looks like unfeasible
<viric>
wpwrak: Ah, page 35 of the PDF has an example of "adapting to an arbitrary load of the line"
<viric>
wpwrak: you need to add a transmission line (or equivalent) with the impedance there written: Z_0' = sqrt(R_L * Z_0) = sqrt(1M * 50)
<wpwrak>
add a transmission line. interesting. let's see what happens ...
<viric>
It has to be a transmission line of lambda\4
<viric>
lambda/4
<viric>
(here the frequency of the signal comes in)
<viric>
So, L = lambda/4, and impedance of the line = 7071 Ohm
<viric>
what does qucs say?
<wpwrak>
hmm, at 10 MHz that would be 5 m :)
<viric>
Can't believe
<viric>
ah. maybe.
<viric>
27Mhz = 11m waves
<viric>
wpwrak: try in the simulator
<wpwrak>
and the signal looks even unhealthier than before
<viric>
oh.
<viric>
can you know the transmission line model of qucs?
<viric>
is the signal narrowband?
<viric>
around a conveying wave? I don't renember the usual english name
<wpwrak>
it gets better if i reduce it to 1 m. but then the first few periods are heavily attenuated. so i guess it's once again something that only works for periodic signals
<wpwrak>
(model) uh, there is the source ... but the math will probably be well above my head ;)
<viric>
ok
<wpwrak>
the signal is DC-~6 MHz
<viric>
there are adaptors that will work perfectly at a fixed periodic wave
<wpwrak>
square wave
<wpwrak>
oh yes, periodic is very easy :)
<viric>
Square wave does not have a very good spectrum :)
<viric>
So, it means you have a broadband signal
<viric>
From 0 to 6MHz as minimum
<wpwrak>
yup
<wpwrak>
i have a comparator i could use to restore a signal that went though a resistive divider, but .... the comparator only works up to ~ 2 MHz.
<viric>
ok
<viric>
If you had a 50Ohm load, you think it allows broadband ?
<viric>
I'd say it also does not allow that.
<wpwrak>
hm, why not ?
<viric>
grr my notes don't have anything about bandwidth. It's all thought for a carrier signal.
<viric>
(I guess)
<wpwrak>
what i'm dealing with here are digital signals. it's for a frequency counter. (ben-wpan/cntr)
<viric>
I think the equivalent of the circuit is an L-C-R circuit as what you saw, with L and C depending on the length of the line
<wpwrak>
so i have a relatively low-impedance source, then my 50 Z transmission line, and the MCU's very high-Z counter input
<viric>
so the length of the line, even if you have an adapted load, is important in all that.
<viric>
(maybe not :D)
<wpwrak>
if you match the impedance, the line length shouldn't matter
<wpwrak>
it's only if you don't have a match that the line length shows up in the form of nice little reflections :)
<viric>
yes.
<viric>
so, you see that I have little idea. Sorry! :)
<viric>
for me, your signal looks too broadband to get all working.
<viric>
I even don't think you can consider the load impedance resistive
<wolfspraul>
wpwrak: I've finished the boomification of the jtag-serial cable for now
<wolfspraul>
it works, generates a digikey shopping list
<wolfspraul>
if you have any feedback let me know, I probably bastardized your beautiful system quite badly :-)
<wolfspraul>
everything but resistors and basic capacitors is just directly linked to digikey order numbers
<wolfspraul>
but still, I think I get the whole system now and looking forward to slowly improving my usage of it
<wpwrak>
(directly link to dk) that's the way i do it too :)
<wolfspraul>
I do not plan to fpedize the one module in there right now, so many other priorities. maybe a bit later.
<wolfspraul>
also I added a little BOOKSHELF file, seems nice
<wolfspraul>
so I will probably move to generating gerbers and files for smt next, maybe not immediately but soon I think
<wolfspraul>
right now the gerbers are committed into git, not sure whether that's a good idea
<wolfspraul>
if the server could automatically create such 'fabrication outputs' and archive them somehow that would be neat
<wolfspraul>
well anyway, boom bottom line: GREAT! works today, and I see huge potential
<wpwrak>
(works) congratulations !
<wolfspraul>
with all the things I learnt with the cable, I feel ready to jump into a bigger project like xue and help there, boomify stuff, clean kicad fields, extend fully characterized matching to more types, etc.
<wpwrak>
kewl ;-)
<wolfspraul>
yeah it's really cool. just make clean show KITS=50
<wpwrak>
btw, you can avoid/reduce the orgy of .sub value changes
<wolfspraul>
well I know I can make it all a lot smoother, it was just an exercise...
<wpwrak>
the way to do it would be to define an equvalence for the vendor part to the digi-key part, with the same part number as in the schematics. that way, you don't need to go via the "private" equivalence and you don't need to tweak the value
<wolfspraul>
you mean the value?
<wolfspraul>
sure the only thing I do now is to add a UJS_ prefix
<wpwrak>
(make clena) or copy over the "again" target that i have in my BOM Makefiles
<wolfspraul>
the important thing is to add more types of parts, more distis, more manufacturers
<wolfspraul>
so I was just playing this time...
<wolfspraul>
anyway, it works! congratulations!
<wolfspraul>
I see a bright future for boom :-)
<wpwrak>
kewl. just generated the BOM too :)
<viric>
what does bom mean?
<wpwrak>
it didn't pick anything overly crazy. the 4.7 uF cap is a bit cheap (Y5V), but usually acceptable
<wolfspraul>
he
<wolfspraul>
I even bypassed the 'exercise' with the 250V cap
<wolfspraul>
just a direct link, lazy me...
<wpwrak>
yeah :) the challenge of getting deeper into substitutions still lies ahead of you :)
<wolfspraul>
ah I have a question about that 4.7nF/250V cap
<wolfspraul>
well the substitutions as a programmatic challenge are totally no problem to me
<wolfspraul>
the problem is to understand what the hell I am dealing with
<wolfspraul>
to me these are all just 'variables'
<wpwrak>
;-))
<wolfspraul>
but without some EE background it makes no sense for me to juggle with these variables
<wolfspraul>
I would just generate bloat.
<wolfspraul>
the point is to not over-specify
<wpwrak>
what i still need to add is an more straightforward debug mode for .sub substitutions. for .gen, i have one, but not for .sub. and it's kinda painful to manually trim the bom for testing
<wpwrak>
yes, once things are a bit more stable and friendly, it'll be a good time to get some input from people who know something about EE :)
<wolfspraul>
so for the 4.7nF/250V... I found these two
<wpwrak>
hmm. don't know really. none of the main differences between c0g and x7r matter here
<wpwrak>
here, you'd be more concerned with overvoltake tolerance
<wpwrak>
overvoltaGe
<wolfspraul>
you mean the 250V? but it's the same
<wpwrak>
i mean what happens if you go over 250 V :)
<wpwrak>
(briefly)
<wolfspraul>
btw haven't tried it yet. can I upload that prettified shopping list to digikey?
<wolfspraul>
do they have some standard format?
<wpwrak>
btw, you could use this one: 490-3532-1-ND
<wpwrak>
smaller and cheaper
<wpwrak>
they have some bom upload but i haven't used it yet
<wpwrak>
if the pretty list was compatible with it, that would be extreme coincidence :)
<wpwrak>
or if you want 1206, try 445-4484-1-ND. that one is even good for up to 630 V :-)
<wolfspraul>
tolerance 20%, is that still oK?
<wpwrak>
of course, if you teach boom to recognize *F/*V, then you don't even need me to look up parts ;-) hint: tell boom that the voltage is >= the specified value
<wolfspraul>
yes yes, I know :-)
<wpwrak>
i don't think the exact capacitance matters here. i think 4.7 nF is very very generous already
<wolfspraul>
OK but I won't touch it, no need to get burnt.
<wpwrak>
oh, and you can get 10% for the same price :) 445-2290-1-ND
<wolfspraul>
so with this little boom lesson, I think the next target for boom for me is xue
<wolfspraul>
for ben-wpan, I'm sure you will do it yourself already
<wolfspraul>
for xue, I believe I can help move things in that direction
<wpwrak>
(ben-wpan) of course :)
<wolfspraul>
wpwrak: how do you feel about committing gerber files into git?
<wolfspraul>
(different subject)
<wpwrak>
it's ugly
<wpwrak>
there is a more general problem, though: how do we deal with generated files that are hard to generate (but that may also be troublesome to have in git, e.g., because of size)
<wpwrak>
one approach would be to have two projects: one with the sources, one with generated stuff
<wpwrak>
that's kinda how i run things with the case scans
<wpwrak>
i haven't quite figured out a strategy for the boom database, where we'll have the same problem
<wpwrak>
making generating gerber easier would be nice. since it's in the same plot dialog as postscript, anyone who makes that possible could also solve the postscript problem -> one step closer to brdhist :)
<wpwrak>
so if you feel like visiting the sewers of C++ ... :)
<Guest5862>
wpwrak: ah yes I see it (the dialog in pcbnew)
<Guest5862>
big :-)
<Guest5862>
why am I called Guest...? :-)
<wpwrak>
big and deep
<Guest5862>
deep? even sub-dialogs?
<B_Lizzard>
Shouldn't devices exist in /sys regardless of their status in /dev?
<Guest5862>
"Save Options" in a dialog, argh. Is that what I think it is?
<wpwrak>
i meant deep in the code. not something easily called from the outside.
<wpwrak>
Guest5862: (save options) don't worry, it doens't really work anyway ;-)
<Guest5862>
he. without the undue burden of trying to send patches upstream, all code paths shall be shallow :-)
<Guest5862>
even in the C++ mess, a few globals can do wonders
<wpwrak>
:-)
<wpwrak>
have you looked at the code yet ?
<Guest5862>
no
<wpwrak>
gets the pop corn
<Guest5862>
but I am very very optimistic on that end. you don't know how cruel I can be to software aesthetics.
<Guest5862>
anyway we see. I'm not sure when I can get to it but automating server-side generated files is definitely something I want to do soon.
<Guest5862>
so in general you say the goal should be to not commit gerbers, at least not in the primary design repository
<wpwrak>
having command-line driven output would also make it much easier to generate local files
<wpwrak>
plus, if you can tweak the settings inthe process, also useful things like mirrored front sides for toner transfer could be automated
<wpwrak>
also because they'd add a lot of commit noise
<wpwrak>
and if you change something but don't generate gerbers afterwards, you may even commit stale versions
<wpwrak>
and so on. lots of problems
<B_Lizzard>
Thanks for that, larsc
<wpwrak>
viric: what probably works is an NPN transistor. Rbase = 50 R :)
<viric>
wpwrak: ah, good one :)
<viric>
but where those 50R come from?
<wpwrak>
i mean: i put a 50 R resistor there :)
<viric>
ahh
<viric>
did you simulate?
<viric>
the base will have a very small current
<viric>
and that may end up being a weird impedance then
<viric>
short-circuit or open-circuit reflect a lot
<viric>
unless Rbase is parallel with the transistor
<wpwrak>
the base will draw whatever it gets
<wpwrak>
Rbase is in serie
<viric>
ah, sorry. I think I was imaging MOSFET.
<wpwrak>
it's an BJT, not a FET :)
<viric>
sure.
<wpwrak>
yup :)
<viric>
so the simulation looks good?
<wpwrak>
yup. very good
<viric>
great... so it really has wide bandwidth once adapted :)
<viric>
I have to learn that qucs
<viric>
I'll package it
<wpwrak>
i need a non-zero capacitance at the input or noise could make the transistor charge the parasitic capacitance at the collector. if course, parasitic capacitance at the base takes care of this anyway.
<viric>
you draw a common emmiter ?
<wpwrak>
qucs is great. you'll love it
<wpwrak>
common emitter ? you mean NPN.E = GND ? yes
<viric>
you brought me into kicad already...
<viric>
yes.
<viric>
does it need freehdl?
<wpwrak>
it says it can read vhdl and verilog
<viric>
ok
<viric>
I'll try to build it
<viric>
wpwrak: is qucs made for qt3? I can't believe... but its configure script looks like wanting it.
<wpwrak>
ldd says libqt-mt.so.3
<viric>
clear.
<viric>
here I go building then.
<lisandropm>
viric: you have caneda, wich is a kind of qucs fork, done with qt4
<lisandropm>
currently it needs some more details, but upstream is alive ;)
<viric>
ohh
<viric>
link?
<lisandropm>
just a second
<viric>
caneda.sf.net I see
<lisandropm>
must fill the IT in debian too
<viric>
no files released...
<lisandropm>
*itp
<viric>
there are no commits in the caneda.sf.net git
<wpwrak>
lisandropm: (caneda) interesting. why did they fork ?
<viric>
ok
<viric>
lisandropm: the qucs mainpage mentions qucs-qt4 in cvs
<lisandropm>
afaik, original upstream was very conservative on multiple things, like keeping autotools for building qt
<lisandropm>
viric: the same guy who started the port went later into caneda (again, afaik)
<lisandropm>
it seems they are integrating µelectronics in caneda too
<wpwrak>
i see layout. kewl :)
<viric>
ok
<viric>
lisandropm: caneda has no released versions
<lisandropm>
no :-(
<lisandropm>
wip ;)
<viric>
usable?
<lisandropm>
still haven't tested it too much
<wpwrak>
it's a little odd that the caneda page doesn't mention qucs.
<lisandropm>
I'm not much into designs :-/
<lisandropm>
wpwrak: indeed
<viric>
qucs works here already.
<lisandropm>
then keep qucs ;)
<qi-bot>
[commit] Werner Almesberger: Added (partially incorrect, see INFO) footprint of F MMCX board edge connector http://qi-hw.com/p/ben-wpan/6668a6d
<qi-bot>
[commit] Werner Almesberger: cntr/term/term.sch: simulation of impedance-matched termination circuit http://qi-hw.com/p/ben-wpan/4176541
<B_Lizzard>
Trying to cat /sys/class/power_supply/voltage_now says "Connection timed out"
<kristianpaul>
lisandropm: www.jlime.com (may be the answer to your problems ;))
<kristianpaul>
cause deian in the nanonote is kinda slow i think
<lisandropm>
kristianpaul: didn't know about it. I just used OE for the devices we develop at my job (because of inheritance)
<larsc>
B_Lizzard: could you do a `cat /proc/interrupts`?
<B_Lizzard>
Am I looking for something or do you want the entire file?
<B_Lizzard>
I'll also give you a copy of dmesg
<larsc>
the entire file would be good
<B_Lizzard>
Sure thing sir
<B_Lizzard>
I still don't get any messages at boot aside from the initial kernel printks, also. Could be my environment, but it worked for 2.6.32 so who knows.
<kristianpaul>
lisandropm: good !
<kristianpaul>
lisandropm: try it and tell us what youn think, btw there is #jlime channel too
<lisandropm>
kristianpaul: sorry, but at this time it would be impossible :-(
<B_Lizzard>
There's some weird output there too
<B_Lizzard>
ALSA is also doing some weird stuff
<larsc>
B_Lizzard: hm thats interesting the battery generates interrupts, but the adc reading still times out
<lisandropm>
viric: no surprise. Most of (if not all) the developers use Debian in their machines, and we have Qt 4.6 (4.7 is not even relaed yet)
<lisandropm>
*released
<kristianpaul>
ok
<larsc>
hm... "Unbalanced enable for IRQ 185"
<viric>
lisandropm: qt47 is not relased1?
<viric>
!?
<lisandropm>
not afaik ;)
<lisandropm>
checks
<viric>
I think it is
<viric>
at least from one or two months
<B_Lizzard>
This is upstream 2.6.36 with those patches and the defconfig I got from projects.qi-hardware
<lisandropm>
iric: sorry, I probably got confused with
<larsc>
B_Lizzard: could you build your kernel with KALLSYMS_ALL=y so we see the function names in the stack trace?
<lisandropm>
4.7.1
<B_Lizzard>
larsc, sure
<B_Lizzard>
Would that be CONFIG_KALLSYMS?
<B_Lizzard>
OK, rebuilding, should be 10 minutes or so.
<larsc>
CONFIG_KALLSYMSÂ Â and CONFIG_KALLSYMS_ALL
<B_Lizzard>
OK, thanks
<larsc>
well, maybe CONFIG_KALLSYMS is enough
<larsc>
yes, i think its enough
<B_Lizzard>
I used your previous defconfig which i modified for CONFIG_BATTERY_JZ4740 and it worked, as far as I can remember
<B_Lizzard>
Could be something you removed?
<B_Lizzard>
Unless this is something that takes time to surface
<viric>
lisandropm: do you use caneda?  I'll try to guess where to install the components. I can't find it in the docs
<lisandropm>
viric: no, but ask Pablo Pareja Obregón parejaobregon at gmail.com
<B_Lizzard>
I did reboot several times and it was the same thing every time
<lisandropm>
while I am an electronic enginnerr, I prefer to do sfotware stuff ;)
<lisandropm>
*software
<viric>
ah found it. It's the first thing it complained about, running it ;)
<larsc>
B_Lizzard: i don't think that it is realted to the other config changes
<B_Lizzard>
Well, me neither, I thought I'd just mention it.
<viric>
lisandropm: you prefer doing software with electronics, over doing electronics with software, I see
<B_Lizzard>
ALSA is giving me some crap too, volume changes in alsamixer don't seem to affect the actual output volume, /dev/dsp is invalid at 14:3 (which I think is the correct major-minor address
<viric>
B_Lizzard: do you have the sound part as module or in kernel?
<viric>
I remember seeing a difference in that
<B_Lizzard>
I think it's in kernel
<B_Lizzard>
Wait
<viric>
but I don't remember what difference.
<viric>
(I talk about jz-2.6.35)
<B_Lizzard>
Yeah, in kernel, as specified in the 2.6.36 defconfig from projects.qi-hardware
<lisandropm>
viric: good :) may I ask for wich distro?
<viric>
NixOS
<lisandropm>
first time I read babout it :)
<larsc>
B_Lizzard: hm, the symbols dont match the addresses anymore, could you build a kernel form the defconfig and paste System.map then?
<B_Lizzard>
Oh, that's from the KALLSYMS build
<B_Lizzard>
The previous pastebins are from the previous build
<B_Lizzard>
Sorry bout that, wait a sec.
<B_Lizzard>
I'll paste dmesg and interrupts when I get a backtrace
<larsc>
ok
<B_Lizzard>
Sorry bout that, I accidentally deleted my sources before :/
<B_Lizzard>
Any ideas on the ALSA issues?
<larsc>
nope
<B_Lizzard>
Shouldn't 14:3 be right for /dev/dsp?
<larsc>
i have no idea
<viric>
on x86 I have 14:3
<larsc>
but isn't /dev/dsp oss?
<viric>
yes
<B_Lizzard>
It should be provided by alsa's OSS emulation
<larsc>
it's not enabled in the defconfig
<B_Lizzard>
Ugh, maybe I'm a moron. CONFIG_SOUND_OSS_CORE is off in defconfig
<B_Lizzard>
I was pretty sure it was on
<B_Lizzard>
That would be CONFIG_SND_PCM_OSS
<B_Lizzard>
The console keymap is missing the red and blue keys too
<larsc>
yes. thats certainly something that should be handeld by userspace and not by the kernel
<B_Lizzard>
Well, aren't good defaults important?
<B_Lizzard>
As the hardware layout is unlikely to change
<B_Lizzard>
Otherwise I see no issues, great work on 2.6.36
<larsc>
well, changing the default kernel keymap wont make it upstream
<B_Lizzard>
Oh, I thought you could do per-target keymaps
<B_Lizzard>
Sorry, you are completely right then.
<viric>
really? aren't there keymaps per target?
<viric>
The macintosh, the alpha, ... they have different keyboards, isn't it?
<B_Lizzard>
Well, I suspect that the base keymaps are the same
<viric>
strange. I had said that different platforms have different scancodes
<viric>
grr my 2.6.35 just hanged in the nanonote. after seeing p1 in mmcblk0
<viric>
(I connected the usb cable booting). The cursor stopped blinking
<viric>
I could reproduce it doing the same.
<viric>
without connecting the cable on boot, it works fine
<B_Lizzard>
Are the limits for sound devices specified in-kernel?
<B_Lizzard>
I suspect they are
<B_Lizzard>
Or are these things hardware-dependent?
<larsc>
limits?
<B_Lizzard>
Um, the steps
<B_Lizzard>
The limits for the Master control are 0 - 3, meaning that each step is a 33% in percentage
<B_Lizzard>
0 - 3 is too low, I think.
<larsc>
thats all the hardware supports
<larsc>
for finer granularity you need a software volume
<B_Lizzard>
OK, thanks
<B_Lizzard>
I still haven't gotten bad driver behaviour...
<B_Lizzard>
It might've been that particular build.
<lisandropm>
viric: my friend just updated the description in caneda's webpage
<lisandropm>
I think it is more accurate now :)
<lisandropm>
wpwrak: also updated they relationship to qucs.
<wpwrak>
lisandropm: nice, thanks ! so this is something that may one day compete with kicad ?
<lisandropm>
seems the idea, as I can read :-/
<wpwrak>
heh :)
<lisandropm>
but it wouldn't suprise me if it takes a more µelectronics direction
<wpwrak>
the layout in the screenshots points in this direction
<wpwrak>
not that i'd mind to have something with nicer selection to compete with eeschema. pcbnew will be harder to beat, though. there's a fair amount of magic that went into all those "magnetic" and "trace-hugging" functions
<lisandropm>
indeed, kicad has a lot of work inside it.
<lisandropm>
but maybe a in-the-middle approach would suit betetr
<lisandropm>
like having something to run simulations and the export to kicad
<lisandropm>
that was my idea when I tried qucs :)
<wpwrak>
yes, that would be useful, too
<wpwrak>
that's basically how i use it. just that export/import is manual :)
<lisandropm>
:)
<wpwrak>
well, in many cases, that's fine. e.g., schematics and simualation of the same circuit may look very different. add a cap here for parasitic capacitance, replace an MCU with a resistor for input port leakage, etc.
<lisandropm>
yes
<wpwrak>
but the basic gui interactions of kicad have a number of design concepts that just don't work well. like the weird way to do block ops, the clumsy concept of proximity in pcbnew, or the ordering of choices in ambiguous selections.
<wpwrak>
alas, so far all attempts to improve this seem to have failed :-( in no small part because each would require some major rearranging of internals to work smoothly.
<lisandropm>
:-(
<wpwrak>
(or so i've read. have't tried my luck with these myself :)
<kristianpaul>
:p
<kristianpaul>
wpwrak: but you you still using geda isnt?
<wpwrak>
and of course, C++ doesn't help with keeping patches small ...
<wpwrak>
kristianpaul: hell, no. i use kicad.
<wpwrak>
kristianpaul: i just see a lot of areas for improvements. e.g., when you click on something in pcbnew, the filtering of objects you may have clicked on is not very good.
<kristianpaul>
wpwrak: ahh , i tought may be i confused u with other people
<wpwrak>
kristianpaul: for instance, it sometimes includes objects that are nearby. with the ability to just zoom in, you don't need this.
<wpwrak>
kristianpaul: then, if there's an ambiguity, it pops up a selection. that's an improvement, because it make a more or less arbitrary pick before. but ... the selection is often redundant. e.g., if i'm about to delete a trace, it doesn't really matter on which segment i click. so if i hit an area where >= 2 segments could be selected, it could notice that they both have the same effect and not ask. this happens very very often.
<kristianpaul>
argg, ouchh
<kristianpaul>
yeah :(
<wpwrak>
kristianpaul: then, if object A overlaps object B at the point we click on, but A is by nature larger than B (e.g., A is a component while B is a trace), then it could give just priority to B, without looking at A.
<wpwrak>
kristianpaul: then, all block operations are by rectangle. there's no way to add/remove items from a selection or to combine selections. one consequence of this is that you now need a menu that lets you select what type of things you really wanted to include, because you often have some "bycatch".
<wpwrak>
kristianpaul: such a menu would be useful in general, but it would be a lot less necessary if selections were more flexible.
<wpwrak>
kristianpaul: (selection) i experimented a bit with such things in fped, that's where all those ideas come from :)
<kristianpaul>
:)
<kristianpaul>
when you decided write fped?
<wpwrak>
(of course, i managed to make it so complex that it sometimes gets confused :) should really take that code apart and redesign it from scratch)
<kristianpaul>
hmm
<wpwrak>
hmm, that was around the time when i looked at some bga, thought about drawing it with a perl script and then reviewing it with kicad's footprint editot, and then very quickly realized "no way !" :-)
<kristianpaul>
wpwrak: are you close with kicad developer, i wonder what you tried with then when openmoko times?
<kristianpaul>
ie sending patches
<wpwrak>
i don't think i contributed anything to kicad when at openmoko. i didn't have much to do with EE back then anyway. well, not officially :)
<wpwrak>
just before starting with openmoko, i sent some patches for "hugging" traces. they were good enough to look exciting but broken enough to still need a month or two of work. luckily, someone picked them up and fixed them for me ;-)
<wpwrak>
since then, kicad development has taken a turn i don't like so much. the code is better structured by the coding style has become incredibly ugly, in my opinion. basically the diametral opposite of the linux kernel coding style.
<kristianpaul>
:-|
<rafa>
lisandropm: btw, your QAntenna software looks great :).. well, I am not a hardware guy, but I just was reading the qantenna README and web page ;)
<wpwrak>
oh. interesting.
<kristianpaul>
yeah
<kristianpaul>
looks really interesting
<wpwrak>
lisandropm: what antenna shapes can it analyze ? e.g., can it do pcb antennas ?
<kristianpaul>
wpwrak: i loads a .net file
<kristianpaul>
whatever it is
<kristianpaul>
lisandropm: what about gps atenas?, i guess nothing to do here cause is focused on active signals on just listen..
<wpwrak>
lisandropm: e.g., i have the interesting problem of "porting" a meander designed for 1 mm FR4 to 0.8 mm FR4. all i know is that i should make it a bit larger, but without much of a clue how large.
<kristianpaul>
wpwrak: how do calculate timing for atusd, are you reading processor registers or just hardconding by guessing that processor always will run to NN Mhz ?
<kristianpaul>
i ask that cause i dont know if by default openwrt modify clock speed.. :-/
<wpwrak>
kristianpaul: i measured the fastest bitbanging speed for a single gpio and i got about 1 MHz. from that, i concluded that i'll never have the slightest risk of going too fast :) (the interface can do 8 MHz, if memory serves me)
<kristianpaul>
ahh
<kristianpaul>
well
<wpwrak>
kristianpaul: you have a SIE, don't you ?
<kristianpaul>
wpwrak: i do
<kristianpaul>
s/do/have
<wpwrak>
good. then you can make the SPI to SPI gateway with the FPGA ;-)
<kristianpaul>
well yes, actually i have a cpld for that job
<kristianpaul>
is more easy to wire and have better looking :)
<wpwrak>
have you already built it ?
<kristianpaul>
nope i just made my jtag cable yday
<kristianpaul>
but i was reading theory about 3 state buffer with vreilog
<kristianpaul>
and i need it for the EVB so i'll with it for sure
<wpwrak>
cool. should be fun :)
<wpwrak>
can't wait for sebastien to do his thesis on bitstream reverse-engineering and open synthesis :)
<wpwrak>
kristianpaul: yes, but they don't handle many of the newer/smaller chips
<kristianpaul>
well open synthesis icarus verilog guys did some work but quit as excuse xilinx have free tools  so no bother?..
<kristianpaul>
wpwrak: who can handle
<kristianpaul>
there are a lot
<kristianpaul>
and more coming
<wpwrak>
sebastien says that icarus isn't really usable
<kristianpaul>
heh he said that for almost everything :p
<kristianpaul>
but indeed it have limitations
<wpwrak>
(almost everything) well, that's true :)
<wpwrak>
have you used icarus ?
<kristianpaul>
a bit
<kristianpaul>
some basic simulation for learning verilog and for the ones that work in Milkymist code
<kristianpaul>
you usually end reporting a bug
<wpwrak>
;-) are the bugs getting fixed ?
<kristianpaul>
well from what i reported they said will take care of and recomendme use some custom IFDEF to make the simulation work with icarus ..
<wpwrak>
so did you get something that worked in the end ? will you use icarus to do the spi gateway on the cpld ?
<kristianpaul>
icarus cant do synthesis at that level
<kristianpaul>
wpwrak: well in partially worked
<kristianpaul>
so lets se NO
<kristianpaul>
:(
<kristianpaul>
sebatian tell me how to fix the code in  verilog, but i dint tried yet
<kristianpaul>
sebastien*
<wpwrak>
ah, pity
<wpwrak>
"can't do" = by design or because of bugs ?
<kristianpaul>
what you meant with do?
<wpwrak>
"can't do synthesis at that level"
<kristianpaul>
surelly i can simulate the verilog code, wich is actually no more that 20 lines
<kristianpaul>
yeah, i uderstood for you that icarus will take care of all the tasks that propiertary xilinx tools do to get  bitstream
<kristianpaul>
so the answers, if that if design is not ready for that
<kristianpaul>
s/if/is
<kristianpaul>
s/if/its
<kristianpaul>
sorry
<kristianpaul>
an nb
<kristianpaul>
damn
<kristianpaul>
i read too fast and make mistakes .P
<kristianpaul>
wpwrak: ( can't do) by design bugs and who know what else
<kristianpaul>
there are several steps to achive that bitstream just using free sofware
<kristianpaul>
let me look a mail thread about it
<wpwrak>
so it's possible with free tools. that's good.
<kristianpaul>
not is not
<kristianpaul>
(there are several steps to achive that bitstream just using free sofware) not achieved yet
<wpwrak>
oh :-( i was hopeful for a moment ...
<kristianpaul>
no sorry, i just want meant there is a lot of work to do
<kristianpaul>
by then poeple to involve in that way
<kristianpaul>
for many years
<kristianpaul>
or
<kristianpaul>
if wolfgang plans about making ASIC sucess (other years to wait for it) wen can have out how FPGA's , but that requires a PhD thesis or soemthing may be
<kristianpaul>
but thats not excuse for you wpwrak dont have a FPGA devel board !! :)
<kristianpaul>
"Good approach, but quite a lot of work that puts off most open source "
<kristianpaul>
developers.
<kristianpaul>
so step by step, as some body in asia said :)
<wpwrak>
yeah :)
<kristianpaul>
wpwrak: is not confusinf for you set MASKS in atdusd.c and otherin hex, why not in binary? , i confess i dont know if C can handle binary data definition?..
<wpwrak>
we shall overcome, one day :)
<wpwrak>
C doens't have base 2
<wpwrak>
i could use octal, though (-:C
<kristianpaul>
argg
<kristianpaul>
damm
<wpwrak>
i should grow a goatee
<kristianpaul>
i'll add coments as much as i can, hex for me still confusing to read :-/
<kristianpaul>
"C doens't have base 2" damn why not?? what about a gcc extension? may be..
<kristianpaul>
ok
<wpwrak>
hex isn't that hard to translate in your head ...
<wpwrak>
it gets more messy if you have, say, 32 bit value s...
<kristianpaul>
hmm
<kristianpaul>
yes it varies
<wpwrak>
and bit fields that end at non-nibble boundaries. that sucks.
<kristianpaul>
i just likt think in bits
<kristianpaul>
lol :)
<kristianpaul>
well
<wpwrak>
you'll like f32x. it has bit entry for boundary scan mode :)
<kristianpaul>
lets comment some ccode
<kristianpaul>
hmm i'll like your way of preset some bits as constants
<kristianpaul>
is like programa  microcontroller :) !
<wpwrak>
yeah, it's important to keep things sane :-)
<wpwrak>
hmm does anyone know if there's a way to make a pattern rule in gnu make that's equivalent to  %.foo: %/%.bar  ? (this doesn't work, because gnu make only substitutes the first % in %/%.bar. since this substitution is done after all other expansions, i can't trick it with $(subst ...) or such either)
<wpwrak>
i could use vpath, though that's a bit brute-force ...
<wpwrak>
and of course,  vpath %.foo %/  doesn't work. grmbl.
<wpwrak>
wolfspraul: heya, want some more feedback on the mm-jtag thingy ?
<wolfspraul>
wpwrak: yes sure
<wpwrak>
alright ... first, you don't have values for FBx yet
<wpwrak>
second, shouldn't R16 also have a voltage rating similar to C31 ? it has a 0402 footprint, so it's rated to only around 50 V
<wpwrak>
third, the mini-usb connector symbol is quite ugly. to make it even more weird, it has two unused pins that don't correspond to anything in the footprint either. how about stealing the one from gta02-core or ben-wpan (the latter got it from the former) ?
<wpwrak>
ben-wpan/components/mini_usb_b.lib
<wpwrak>
fifth, (that's something inherited from kicad) you have a nice example for the trouble with the default symbols when you look at the font sizes in FB2, then C27, and finally +3.3V (all are near each other)
<wolfspraul>
what do you mean with 'values for FBx'?
<wpwrak>
finally, in case you try to do a pcbnew --plot/--gerber, maybe consider also eeschema --bom, so that we can remove this manual step as well
<wolfspraul>
you mean the current rating? you want it to be in the KiCad fields?
<wpwrak>
current rating, maybe resistance, or just a part number
<wpwrak>
right now it's a little under-specified ;-)
<wolfspraul>
hmm OK. good feedback thanks!
<dandon>
are multi-touch displays compatible with windows xp?
<wolfspraul>
for the beads, actually I realized I didn't know to label them better, or what a 'good way' would be
<wolfspraul>
wpwrak: for example, would you just put "1A" into the value field?
<wpwrak>
i'm not so sure about them myself. there are two parameters that are interesting: one is the current, the other the DC resistance
<wolfspraul>
he :-) so we are back to where we were 2 days ago. that's why they don't have proper values yet.
<wpwrak>
for now, you could just put the part number you already picked :)
<wolfspraul>
into the value field?
<wpwrak>
yup. get a part number match
<wolfspraul>
hmm
<wolfspraul>
even though the value field shows in the plotted schematics?
<wpwrak>
why not ? it characterizes the part
<wolfspraul>
hmm
<wolfspraul>
that number doesn't make the schematics more readable
<wolfspraul>
then rather maybe "BEAD 1A"?
<wpwrak>
i wouldn't put type names into the value. in principle, the symbol should identify the type.
<wpwrak>
and, for boom, the component reference.
<wolfspraul>
I don't think I will put digi-key part numbers into the schematics now. I let you do the first step, once I see this in projects of yours I follow :-)
<wpwrak>
i was thinking of the manufacturer part number, not distributor part numbers
<wolfspraul>
ok I have no strong feelings on this now, because I work with these documents too little, so I don't know when exactly I look at them and what I then expect from them.
<wpwrak>
and there's precedent. e.g., in simple little cntr, we have U1 and D1, both with vendor PN
<wpwrak>
for U1 it's so natural you don't even notice. for D1 a little less. but then, LEDs have a ton of characteristics ...
<wpwrak>
beads also have their share. e.g., besides the obvious parameters, i think you can get them shielded or not, perhaps with material variations, and so on
<wpwrak>
lisandropm: would you know a rule of thumb for resizing such a meander antenna when the PCB thickness changes ? i.e., the example is for a 1 mm board, but i'm using 0.8 mm.
<lisandropm>
wpwrak: sorry, but no
<lisandropm>
I actually wrote much of the GUI stuff
<lisandropm>
the calculation part was done by necpp people ;)
<lisandropm>
GUI stuff: qantenna ;)
<lisandropm>
should really create xmls or the like for describing antennas in a much clearer way :-/
<wpwrak>
just thought you may have encountered such a problem already, since you know about such things .. :)
<kristianpaul>
lisandropm: i guess poople ask you that all time :)
<lisandropm>
indeed, happens very often
<lisandropm>
or I am approached by people who used nec a lot and discover the gui, and ask me questions that I really don't have a clue of :-)
<kristianpaul>
is nec pople reacheble some how? irc or mail list may be?
<lisandropm>
most people user supernec or 4nec2, wich are propietary
<rafa>
kristianpaul: we tried ffmpeg2theora before, but it did not gave us good videos to see
<kristianpaul>
i need a small movie
<kristianpaul>
later ogv
<xiangfu>
wolfspraul: I think I fixed the 16gb sd card read bug. please test with your sd card. :)
<kristianpaul>
converting some personal video now btw
<xiangfu>
wolfspraul: I have send one email to lists. you can find the new u-boot image at that email :)
<wolfspraul>
cool!
<rafa>
kristianpaul: I found that for some videos mplayer shows the first frame for several seconds (maybe 10) until it continues okey.. so if you see this behaviour that is okey. I mean, it is the same thing I get.. no idea why. Maybe it is mplayer, maybe it is that video, maybe it is theora.. who knows
<kristianpaul>
i think mplayert take a while to load the video
<kristianpaul>
ok i'll inform you, as soon this finish converting