Topic for #qi-hardware is now Copyleft hardware - http://qi-hardware.com | hardware hackers join here to discuss Ben NanoNote, atben / atusb 802.15.4 wireless, and other community driven hw projects | public logging at http://en.qi-hardware.com/irclogs
xiangfu has joined #qi-hardware
* wpwrak
wonders if rjeffries in return sings the praise of qi-hw on the raspberry pi channel
<wolfspraul>
does it matter?
<wolfspraul>
the raspberry pi has pretty much nothing of interest, but bad art has never necessarily stopped people from lining up for it :-)
<wolfspraul>
a broadcom marketing gimmick, oh my :-)
<wolfspraul>
did anybody see the success of the pebble watch on kickstarter?
<wolfspraul>
that's something new
<wolfspraul>
they raised over 5 million USD in preorders for a new product, cool!
<wolfspraul>
that will kick some of the lazier VCs into shape to provide real value to their startups
zedstar has joined #qi-hardware
zedstar has joined #qi-hardware
<wolfspraul>
it's not open hardware or anything, but a great success on financing through preorders. their timeline and history suggests that they can actually deliver... cool I think
<rz2k>
I actually dont get why rpi guys cant make manufacturing faster, with 200k+ preorders...
<wolfspraul>
well
<wolfspraul>
what is the rpi actually?
<wolfspraul>
think about it
<rz2k>
broadcom sponsored company for world domination :3
<wpwrak>
(watch) yeah, pretty impressive
<wolfspraul>
it's broadcom's answer to the 'successful' beagleboard, and on the other side also something to sneak developers from qualcomm/atheros
<wolfspraul>
a sad thing, marketing scheming all around
<wolfspraul>
I feel sorry for people that waste their time on this, but ok
<wolfspraul>
there are lots of hobbies one can have :-)
<wolfspraul>
so we innovate on things like mechanical, sourcing, testing, eda tools, etc.
<wolfspraul>
there is actually zero overlap to something either like the beagle* or pi
<wolfspraul>
the overlap is with projects like elphel, osmo*, arduino, maybe some others. most make pretty bad compromises on tools though, seems they don't see much value in a free process. so we try here :-)
<wolfspraul>
I hope we accelerate with m1
<wolfspraul>
transfer to kicad, then source with boom finally. maybe even layout over, why not.
<wolfspraul>
want to speed up and leave the legacy tools behind us so we can be faster and cheaper on future hardware
<wolfspraul>
oh and I hope the migen switch actually happens
<wolfspraul>
and gives us faster memory and then higher resolution, while the board moves from analog video-out to digital video-out
<wpwrak>
more pixel depth ! :)
<wolfspraul>
also, sure
<wpwrak>
i'm getting very tired of staring at all that green ;-)
<wolfspraul>
with our (slow and good) speed, by the time we are done the master schemers have already dueled a bit more and there will be less options
<wolfspraul>
I understand broadcom is fighting, they should
<wolfspraul>
unlike some others who are already 'done', broadcom still has a chance to maintain their independence
<wolfspraul>
so management should work hard to keep their good paychecks coming :-)
<wpwrak>
and they're moving in the direction of openness, which is always a good thing
<wolfspraul>
where do you see that?
<wpwrak>
e.g., they released fairly detailed soc documentation. that wasn't around before.
<wpwrak>
and we've seen companies even go in the opposite direction
<wolfspraul>
everybody is competing for limited developer talent
<wpwrak>
it's also nice that they're releasing their schematics, at least (nice) at a symbolic level
<wolfspraul>
but sure, every chip with an open datasheet is a good chip
<wolfspraul>
rz2k: what do you plan to do with the beaglebone?
<wpwrak>
i prefer companies to compete on openness than on who has the prettier shackles ;-)
<wpwrak>
have you introduced him to MIDI on the M1 yet ? :)
<wolfspraul>
good point, not yet
<wolfspraul>
rz2k: please keep us posted about your beaglebone endeavor
<wpwrak>
hah, gotcha ! :)
<rz2k>
ok :)
<wolfspraul>
I understand that many people feel this urge to move past the boring desktop/notebook world, and it's tough to find a good starting point in the 'beyond' world
<wolfspraul>
any step is good, as long as you learn and quickly adjust so that you get the most satisfaction out of it
<wolfspraul>
I don't have anything 'better' than the beaglebone or pi for you either
<wolfspraul>
the stuff we work on here is also pretty exotic and requires you to be in a drugged state, more or less, to enjoy :-)
<wolfspraul>
though one you get the hang I think it's pretty cool and rewarding
<wolfspraul>
oh btw, I saw this Ikea & TCL teaming up news. Jon and I are working on some connections with large furniture makers
<wolfspraul>
maybe with Ikea making such a highly public move, something interesting may emerge there
<wolfspraul>
I think Ikea has a great sales channel to sell goods that include embedded electronics
<wolfspraul>
Best Buy is closing stores (thank god), I don't think Ikea is :-)
<wolfspraul>
ran into the section "Is this just a TI marketing gimmick?"
<wolfspraul>
ha ha
<wolfspraul>
I see they listened a bit better in the propaganda 101 class
<rz2k>
yeah, they tried to confront that, but we know the truth :3
<wolfspraul>
but ... with me living in Beijing ... *NOTHING* can beat the masters I see working at this game every day around me
<rz2k>
I'm sure circuitco is funded by TI.
<wolfspraul>
the guys here in the capital are so good at it, they will make your head hurt with all their twists and turns in manipulating people :-)
<wolfspraul>
that TI & Broadcom stuff looks like childs play in comparison
<wolfspraul>
anyway
<wolfspraul>
so no, it's not a TI marketing gimmick
<wolfspraul>
of course not
<wolfspraul>
they even answered it already, ahem
<wolfspraul>
:-)
<wolfspraul>
it seems the beaglebone is more or less directly aimed at arduino
<wolfspraul>
which already is quite an eye opener since for any real open project, there is no competition in the old sense of who owns what.
<wolfspraul>
say for our work here, if some other great 'open' project comes along. oh my, we are so so scared.
<wolfspraul>
really
<wolfspraul>
oh god, that must be the end
<wolfspraul>
ah no wait, maybe not?
<rz2k>
check their cloud9 ide thats installed on sd card by default
<rz2k>
its javascript controlling hardware :S
<wolfspraul>
there are exactly 2 options, in each such case:
<wolfspraul>
1) the project is not actually open, then of course it is not interesting and most likely there are far better closed projects out there already
<wolfspraul>
2) the project is open, or parts of it, and then those parts are also *reusable*, by definition
<wolfspraul>
so any successful open project creates opportunities for other open projects
<wolfspraul>
but I think 90% of 'open' projects are not very open beyond a new trick at redefining what open means
<wolfspraul>
which makes them irrelevant in the context of the actual open projects :-)
<wolfspraul>
hope this makes sense :-)
<wolfspraul>
that's so funny that they answer the "marketing gimmick" question themselves there
<wolfspraul>
actually I think if it is a marketing gimmick, you can still try to use it to your advantage as much as possible
<wolfspraul>
probably you should
<wolfspraul>
but watch out, they don't care that much about how well it actually works for you...
<wolfspraul>
there's no room for that in the, ahem, marketing budget :-)
<mth>
so the question is not whether it is marketing, but whether it is a gimmick...
<wolfspraul>
I think many times our discussions about patents, copyright etc. are totally removed from the real world
<wpwrak>
wolfspraul: it's a moving battle :)
<wolfspraul>
once any of this stuff would get to an actual trial, it would be mangled down in horrible ways unimaginable to us anyway
<wolfspraul>
that's not to say that we should let down our quality standards, not at all
<wolfspraul>
but I can painfully imagine how the poor jury and judges and others have to endure these infinite attempts at APIs and what not
<wpwrak>
cladamw: DIODE-SOT-AXC emphasizes the characteristic of being a diode. DS emphasizes the package (3 pins). both are equivalent.
<wolfspraul>
already wondering whether they will be caught up in rush hour traffic jam on the way home, or get out a bit earlier today! :-)
<wpwrak>
cladamw: similar to your 74AUP1G08GW,125 emphasizing the package with my 74X1G08_5 putting slightly more focus on the function
<wpwrak>
wolfspraul: regarding juries, i wonder how a direct attack on the basis of the patent system would work. namely the legal fiction that "infringing" on a patent implies that any value has been obtained from it
<cladamw>
wpwrak, okay ... so just a name with emphasizing 3 pins. but since I'm now need to use 0805 with such same symbol (surely KiCad has it ), so do we keep to use yours or we create another generic naming for diode lib ?
<cladamw>
although we knew the model link we can link to specific one. :)
<wpwrak>
cladamw: you can't use DIODE-SOT-AXC "as is", because it uses pins 1 and 3. but you could make an adapted version for pins 1 and 2, compatible with 0805, etc.
<cladamw>
wpwrak, aha ... i see. since I don't get into seeing its pins assignment, sorry that i asked a stupid question. oaky... now I go to create a generic one.
<wolfspraul>
who knows you know it's all speculation
<wpwrak>
a diode for 0805 (or also any number of through-hole footprints) would be DIODE-AC, if following the same naming convention. the "-SOT" was probably a bad idea, although it helps to clarify why we need the "AXC"
<cladamw>
so I'll give a named 'DIODE-AC' :-O
<wolfspraul>
I think we are reasonably safe from vicious attacks, and respect other peoples work (of course). should be fine.
<wpwrak>
wolfspraul: sure. i just wonder if anyone has actually tried that. this ought to be the weak point of a jury.
<wolfspraul>
I will pay more attention to trademarks too, stay away from trademarks that are being gobbled up with whatever money making scheme.
<wolfspraul>
does anyone know how the performance of a pcb antenna (say wifi or 802.15.4) is affected by a strong magnet nearby?
<wpwrak>
the magnetism doesn't matter. the metal of the magnet probably does
<wolfspraul>
can magnets in general impact the performance of ics, do they have an impact on the electrons flowing around?
<wpwrak>
(assuming a conventional magnet)
<wolfspraul>
ok sure, the metal is clear. but the magnetic field has no impact?
<wpwrak>
normally, very little
<wpwrak>
of course, if we're talking about an MRI or such, things change
<wolfspraul>
MRI?
<wpwrak>
cladamw: DIODE-AC sounds good to me. or simple DIODE. it's the most basic layout anyway :)
<wolfspraul>
I was thinking of magnets used to make things hold onto each other
<cladamw>
wpwrak, yuh. okay.
<wpwrak>
Magnetic resonance imaging
<wpwrak>
oh, these are WEAK ;-)
<wolfspraul>
ok
<wolfspraul>
so you would not even expect an impact on a pcb antenna
<wolfspraul>
great :-)
<wolfspraul>
I was worried the field may just crowd out the spectrum, or whatever. since it's magnetic, the rf is electromagnetic (?) well, I am guessing/wondering
<wpwrak>
the magnet just has a few millitesla at most. and it's fairly static. it's field changes that cause electrons to move.
<wolfspraul>
ah that makes sense
<wpwrak>
strong magnets are in the order of 1 tesla or more. and even then, if the field is static, not much happens
<wolfspraul>
ok but say you plug your electronic device against a metal wall, to attach it there
<wpwrak>
i'd worry a lot about the metal wall :)
<wolfspraul>
then the mechanical action of moving the logic board into the field, or changing the field (?) because it's now holding to the other side could cause a small disturbance?
<wolfspraul>
but yeah, I got it
<wolfspraul>
static field, weak, etc.
<roh>
wolfspraul: i think you are in the clear for regular magnetic forces in a household
<wpwrak>
there will be a small disturbance, yes. not sure if it'll be strong enough to upset reception. in any case, it's very limited in time. while the huge reflecting metal wall will be there - in comparison - forever
<roh>
emi from powertools is as bad as it gets for most 'disturbances of the force' ;)
<wpwrak>
roh: excluding present company, perhaps ;-) oh wait, sebastien isn;t around. then i guess we're safe :) (he sometimes gets ideas about the dark side of physics)
<wpwrak>
yeah, power tools are evil. sparks and all
<xiangfu>
I copy some videos from this factory. not sure about the license. I just uploaded first.
<wpwrak>
soldering ... how not to do it ;-)
<wpwrak>
xiangfu: i didn't know you had entered the film business, directing a horror movie ;-) well, 0201 #1 and #2 have a happy ending, but ...
<wolfspraul>
xiangfu: ok wait. let's find out the license and delete unless it's specifically ok to upload them as public domain or whatever.
<wolfspraul>
these clips look like instructional clips from a smt or reflow maker maybe
<wolfspraul>
I understand that smt factry staff 'gave' them to you, but we need to determine the origin or license, don't want to accumulate content of unclear origins in the qi wiki...
<wolfspraul>
once we accept some sloppiness, I think it's a slippery slope and over time we loose control of the qi content quality altogether
<wolfspraul>
had enough trouble with Carlos students freely copy-pasting a while back :-)
<wpwrak>
立碑 = Erected a monument. nice ;-)
<wolfspraul>
those are nice videos btw, but we need to determine the origin
<xiangfu>
wolfspraul, ok.
<wolfspraul>
we determined that we need to delete the videos until the origin is clear
<wolfspraul>
the guy who gave them to xiangfu 'forgot' where they are from :-)
<wolfspraul>
they look like videos from a pick & place or reflow maker to me
<cladamw>
since my local MiscControl.sch got unexpected loss lots of texts after I filled-in non standard pdf link (i.e. includes non-ASCII code link.) :(
<cladamw>
except using that 'download' icon to download, is there other git cmd, i can use ?
<wpwrak>
if you're looking for non-ASCII, here's some more: MCE-100 and SCJ368R1NUS0B00G both contain things vim highlights. in MCE-100, they're at 5.6mV/Pa, ????9.7
<DocScrutinizer>
wolfspraul: static magnetic fields usually have pretty little impact on electric circuits. It needs specially designed components exploiting exotic effects like Hall-effect or GMR, to even *detect* magnetic fields
<DocScrutinizer>
chokes with ferromagnetic core however can get saturated by static magnetic fields, causing a massive loss of inductance of the choke
<kyak>
does qi hardware have a competitor? is there such a thing as competition in open hardware world?
<kyak>
i often see how you bash arduino or bb or rpi.. but why do you do that? are they not "open" enough? or maybe you feel a competitor?
<rz2k>
bb and rpi arent open at all. :3
Martix has joined #qi-hardware
cladamw has joined #qi-hardware
paroneayea has joined #qi-hardware
<wolfspraul>
kyak: you see me bashing them? I should re-read the chat archives and improve my language then :-)
<wolfspraul>
the arduino project in particular is amazing
<wolfspraul>
I think I told the story how I was wondering around an 8 or more level bookstore in Tokyo once, and felt totally lost among all the Japanese characters
<wolfspraul>
until I finally saw a book about...
<wolfspraul>
ARDUINO :-)
<wolfspraul>
then I knew there must be friends there, somewhere in this foreign city :-)
<wolfspraul>
the pi is also very interesting because it adds hdmi output (and 256 meg ram) at that pricepoint
<wolfspraul>
although so far I was able to solve all problems in that category with openwrt-based routers
<wolfspraul>
I agree that there is no competition in open projects, at best they can help and lift up each other
<wolfspraul>
that's the whole point imho
<wolfspraul>
kyak: what do you think of these projects? [bb/pi/arduino] are you using any of them? what do you like the most?
<kyak>
wolfspraul: i'm using arduino and bb to showcase control systems development in Matlab/Simulink
<kyak>
mostly for academic
<kyak>
Lego NXT is the third one
<wolfspraul>
nice! please tell us more
<kyak>
i'm not sure if you are interested, it's not realted to open source :)
<wolfspraul>
what is the best part of them? (what you like the most)
<wolfspraul>
well
<wolfspraul>
I think it is, actually. the boundary is not clear, right?
<wolfspraul>
and you speak freely, so with your own actions of freely sharing what you do and learnt, you make them more open :-)
<kyak>
we are using Lego and Arduino to show control systems development, and bb to show different image/audio/video processing techniques
<wolfspraul>
you seem to have a really large base actually
<wolfspraul>
you are a Ben user, also arduino, bb, milkymist - wow! :-)
<kyak>
(on the left you can find the same for bb/ lego etc)
<kyak>
so you hook up any of this shit to your laptop, develop some block diagrams in Simulink, and then see it running on real hardware.. that's in a nutshell
<wolfspraul>
what is simulink? (checking)
jekhor has joined #qi-hardware
<kyak>
it is a software for modeling of dynamic systems
<kyak>
the key point is that you can test your design early, and then you are using automatic C/C++ code generation for implementation
<kyak>
so these cheap and available hardware like arduino help us demonstrate many of modern paradigms, like model-based design
<wolfspraul>
how much do you pay for simulink?
<kyak>
if you use it commercially, a lot (well, hard to define "a lot"). I wouldn't afford that. Company would affort it. FOr educational use, it's much0much cheaper
<wolfspraul>
he
<wolfspraul>
I think that will always remain a strange concept to me, I maybe a little old fashioned about 'education'
<kyak>
it becomes very clear why if you think that they are students now, and in 5-10 years they are working for some companies
<wolfspraul>
what are the closest free alternatives you know that could solve a similar problem?
<kyak>
there is nothing like that in open source.
<wolfspraul>
interesting
<kyak>
not even close
<kyak>
there are non-free competitors, but they are also far away
<wolfspraul>
I will read more about it
<wolfspraul>
thanks! always eager to learn
<wolfspraul>
:-)
<kyak>
Matlab/Simulink is a de-facto standard for modeling and simulation and deployment in automotive/aerospace/defense
<kyak>
and btw, Simulink models can run on Ben (just like they run on arduino or bb or TI processors or whatever else): https://github.com/kyak/nanonote_ert
<wolfspraul>
in your projects/work, what comes after simulink?
<wolfspraul>
you first model a dynamic system in it, then ?
<wolfspraul>
how does it go to an arduino or bb?
<kyak>
for open source alternatives, Octave is pretty much the clone of Matlab and Scilab is a clone of Simulink. But again, it's light years behind
<wolfspraul>
it's compiled and linked with libraries, then executed as a program? via hdl it requires an fpga?
<kyak>
then Simulink generates C code from your model and you can feed it into your IDE/toolchain to deploy on the final target
<kyak>
for hdl it requires and fpga or hdl simulator
<wolfspraul>
ok it starts to make sense :-)
<kyak>
it actually makes a lot of sense, and also pays my bills :) I used to work for several automotive companies developing the software for ECU's using Simulink (and i'm not a programmer, i don't have to be to develop in Matlab/Simulink). And now i'm what is called an "Applicatin Engineer". Teaching others how to solve their problems using our tools
<wolfspraul>
what is an ECU?
<kyak>
Electronic Control Unit
<kyak>
i thought all people in Germany know that ;)
<kyak>
or Engine Control Unit, which describes it better
<wolfspraul>
still wondering about scilab and simulink
<wolfspraul>
if the models are simple, at some point they must be simple enough to be representable with scilab, no?
<wolfspraul>
so why wouldn't you use scilab until you reach a feature that is only possible in simulink, and then use simulink for those advanced models?
<wolfspraul>
I would probably want to approach problems bottom-up
<kyak>
it's not the model's simplicity/hardness that matter.. It's a question of your complete workflow,.. Like having requirements linked to your models and back, code generation, intensive early verification and validation, certification artefacts.. you get nothing of this (and many else)
<wolfspraul>
but you are so well versed in free operating systems and tools :-)
<wolfspraul>
if even you say that you can't do it there, that's amazing
<kyak>
i like open source :)
<kyak>
but when i test octave vs matlab and octave is like 100 times slower, what can i do?
<wolfspraul>
I will read more about this, and try out. slowly moving to modeling, digital systems, dynamic systems, etc.
<kyak>
the only 'pros' is that octave actually runs on Ben :)
<wolfspraul>
yeah, amazing
<wolfspraul>
I'm trying to find a reason to use it :-)
Ayla has joined #qi-hardware
<wolfspraul>
can I model signal filters with it?
<kyak>
sure can
<wolfspraul>
I should try
<wolfspraul>
ah ok
<wolfspraul>
good! :-)
<kyak>
yeah, i've shown octave running on ben to my collegues, and they liked that "small Matlab" :) But then they show me their iphones running matlab in a cloud, and i feel like i'm from 90s :)
<wolfspraul>
oh sure, I can imagine
<wolfspraul>
they run matlab on a cloud server, and only the gui on the iphone?
<kyak>
right
<wolfspraul>
nice
<kyak>
and it's "free" btw.. you only got to have a license
<wolfspraul>
well, I see through that definition of 'free', easy to say 'no thanks'
<wolfspraul>
but the usability is great, no doubt
<wolfspraul>
you could model systems and let them run on your milkymist board via hdl
<wolfspraul>
do you have other fpga boards as well?
<kyak>
we have spartan-3a, spartan sp-605 and virtex ml-605 is on it's way..
<kyak>
probably can find something from ALtera as well
<wolfspraul>
he
<kyak>
is there a place for "user" hdl code on milkymist?
<wolfspraul>
what do you mean?
<kyak>
i mean, it has hdl soc and then all you got is C?
<kyak>
it's like another layer
<wolfspraul>
you can synthesize additional hdl next to the soc
<kyak>
ah!
<kyak>
that's exactly what i asked :)
<wolfspraul>
but it's not well documented yet, so someone doing it and documenting the process would be very valuable
<wolfspraul>
kristianpaul has been doing some stuff
<kyak>
does it have a lot of area left from soc (even roughly)?
<wolfspraul>
yes, over 50% I think
<kyak>
so one could make it a "dual core" :)
<wolfspraul>
oh, definitely
<wolfspraul>
especially since the lm32 core is small
panda|x201 has joined #qi-hardware
<wolfspraul>
but it's all work, and nobody on this right now afaik
rejon has joined #qi-hardware
<wolfspraul>
alright, yum install scilab is running :-) I think I am stupid enough that scilab will be enough for me :-)
<kyak>
it's reasonably enough to taste modelling
wolfspra1l has joined #qi-hardware
rejon has joined #qi-hardware
Ayla has joined #qi-hardware
rejon has joined #qi-hardware
infobot has joined #qi-hardware
<wolfspra1l>
kyak: what kind of systems are you typically modeling?
<wpwrak>
hmm, i'd say there is a bit of competition between open hw projects. e.g., arduino and ben should have a little bit of overlap.
<kyak>
wolfspra1l: typical applications are control systems, digital signal processing and embedded systems.. but there are many more applications
<wpwrak>
and all compete for public exposure
<wpwrak>
so it's in part common goals, in part competition among each other. nothing new really. also ipad and the samsung tab share a similar relationship. only that the companies involved have too much money for lawyers :)
<kyak>
wpwrak: but then, arduino doesn't have a fully-fledged operating system, while Ben doesn't provide the number of IO arduino provides.. so it separates immediately target audience in two parts at least
<kyak>
i'm not sure how i can hook up a DC motor to Ben.. and even if i can, how do i control it?
<wpwrak>
kyak: sure, they're not the same. but they have some overlap. if your application is in that overlap area or if your idea of what you want to do is sufficiently fuzzy, both may appear suitable
<wpwrak>
(hooking things up to ben) UBB ;-)
<wolfspra1l>
kyak: thanks [models]
<wpwrak>
remember, we have stable XGA out on something that's basically UBB. beat that, arduino ;-)
<kyak>
hehe
<wpwrak>
at the moment, i use UBB mainly to program microcontrollers. i have ben-based programmers for silabs (8051), avr, and pic18. they're all little more than a UBB plus a bit of cable. well, unless i want to go fancy and make a luxury version with spring-loaded pins.
LunaVorax has joined #qi-hardware
<kyak>
wpwrak: did you try generating a PWM signal with UBB? Is it even possible taking we are running linux?
<wpwrak>
depends a bit on how accurate it has to be. for VGA, i found i had to disable everything else, at the level of effort i put into that (may be possible to let the CPU run a little in the background, but it would need more work)
<wpwrak>
if your timing is less demanding, then you can probably get away with more concurrent activity
<kyak>
what was the timing again you achieved for VGA out?
<wpwrak>
you have distortions at several levels: memory bus contention, cache misses, and then cpu scheduling
<kyak>
so if i want to drive my motor with ~490 Hz PWM signal, i can assume it would work? Or you never know untill you try? :)
<wpwrak>
what resolution does the duty cycle have to have ?
<wpwrak>
in any case, anything within a few kHz isn't very demanding. you'll need to experiment a bit, though. it's not something the chip is designed to do, so you a bit of lateral thinking is needed
<wpwrak>
just like ubb-vga pretends to be an MMC device :)
<kyak>
yes, it should be something within few kHz
<wpwrak>
one thing that limits ubb-vga is that it runs entirely in user space. there are some things you can't do there.
<kyak>
speaking about distortions you mentioned above.. are they also the case if i run my code on Ben in "barebone" mode (i.e. passing control from uboot directly to my code)
<wpwrak>
well, ubb-vga pretty much does a "barebone" mode ;-) i didn't turn off the cache, though
<kyak>
i see
<wpwrak>
i'll need to add a date and hour tag to kicad-libs-components.pdf :)
<wpwrak>
ubb-vga still has access to the full linux functionality before and after its pixel-pushing mode. so, for example, i can load the screen content from a file system
<wpwrak>
by not running in kernel mode (although i shut down interrupts and such), i can't halt the cpu until an interrupt happens. that makes my synchonization with timer a few clock cycles less accurate than would be possible
<wpwrak>
but yes, we're talking about tens of nanonseconds here
<wpwrak>
if you want to have full linux running in parallel, you could arm a timer to fire a bit before the moment you really need it, then busy-wait until the right time. this assumes the jitter (mainly the longest amount of time the kernel is running with interrupts disabled) is below the time between events
<kyak>
ok.. this is pretty good! do i understand correctly that UBB pins (all 8 of them) act as GPIO? So we can make them act as digital inputs or outputs
<cladamw>
wpwrak, yes, add data and hour tag as i'm doing DRC. :-) All err I debugged out excepts 66 warnings. well ... it's late here. Needs to fix them next week. :-)
<wpwrak>
lars also suggested to try making the whole thing dma-based and even do MMC controller operations via DMA. i didn't explore that. it may give you even better control.
<cladamw>
wpwrak, with data and tag then we can track back if need on checking kicad-libs-components.pdf , tks a lot. :-)
<wpwrak>
(ubb) six of them are full GPIOs. all but one have an internal pull-up. plus, you have VDD which you can switch on or off, but only slowly. (there's a big cap on it)
<kyak>
wpwrak: ok, thanks
<wpwrak>
cladamw: i update the catalog one or twice a day. so if you change anything in the evening, you can be reasonably sure there's an updated catalog by the time you wake up again :)
<cladamw>
wpwrak, hi, alright. Nice. :)
<cladamw>
night here. cu
<wpwrak>
kyak: VDD has another quirk: the cap is on the wrong side of the switch, so it adds to the inrush current. that's why it's often necessary to "precharge" external devices.
<wpwrak>
"precharge" = disable VDD, drive all I/Os high. wait ~100 ms. then enable VDD. this way, current will crawl through the I/Os and charge the caps on the device side, including the 10.1 uF inside the ben
<kyak>
wpwrak: so if i don't "precharge", i can't use the GPIOs?
<wpwrak>
if you don't precharge, the ben's 3.3 V supply may drop when you enable power to UBB and the ben freezes
<wpwrak>
it all depends on what exactly is connected there, of course
<wpwrak>
lesson learned for the ya: that switch should have a current-limiter :)
orson_zhai has joined #qi-hardware
qwebirc97987 has joined #qi-hardware
orson_zhai has joined #qi-hardware
<qwebirc97987>
Hello there
<qwebirc97987>
Hi
<qwebirc97987>
I want to buy a miny computer
<qwebirc97987>
something like raspberry pi
<qwebirc97987>
could you help me please?!
infobot has joined #qi-hardware
<mth>
qwebirc97987: what do you want to use it for? and how open must it be?
GNUtoo-desktop has joined #qi-hardware
emeb has joined #qi-hardware
<qwebirc97987>
ok
<qwebirc97987>
I just want to run some little processes all day
<qwebirc97987>
with it
<qwebirc97987>
and I need a LAN on it
<qwebirc97987>
I want to have a GNU/Linux distribution on it
<mth>
does it need its own screen or the ability to connect a screen, or just ethernet?
<kyak>
wpwrak: ok, thanks for explanations :)
<qwebirc97987>
ethernet is enough if I can program it from my pc
<qwebirc97987>
*Sorry for my bad English!*
<qwebirc97987>
*I dont now English very well*
<qwebirc97987>
but it is better to have the ability to connect to a screen
orson_zhai has joined #qi-hardware
<mth>
Beagle Bone was mentioned yesterday, it has ethernet and probably a way to connect an LCD panel (haven't checked that part)
<mth>
Raspberry Pi might be useful, once it starts shipping in large quantities; the video driver is a binary blob though
<mth>
SheevaPlug is network-only
<qwebirc97987>
mth: thank you very much for your help!
<mth>
maybe a wifi/internet router might suit your needs as well; some can be reflashed with Linux
<mth>
check for example OpenWRT supported devices
jekhor has joined #qi-hardware
<Ayla>
I'd say a Raspberry Pi, if you can wait a couple of months
<mth>
two months (tm)? ;)
<qwebirc97987>
Alya: are you sure Raspberry Pi is available for next two months?
<qwebirc97987>
How do tou now that?
<qwebirc97987>
*How do *you* now that?!
LunaVorax has joined #qi-hardware
<mth>
"two months" is a joke about Pandora, who have had trouble actually building and shipping units
<Ayla>
I hope the R-Pi team will handle this better
<viric>
the sheevaplug is quite good
<viric>
noone here uses sheevaplugs?
<viric>
wolfspra1l: could you get 'openssl speed' running in the milkymist?
<Ayla>
viric, I was about to buy one, then I heard about the rasperry pi
<viric>
the r-pi has half or a quarter of the sheevaplug ram
<viric>
ram is of big vlaue :)
<viric>
value
<viric>
Ayla: unfortunately, the sheevaplug and the guruplug have awful power supplies. I replaced the capacitors twice, in it
<viric>
but very compact, at least.
<viric>
it's nice to plug 220V directly to the small computer :)
<viric>
imagine the nanonote with a 220V plug.
<viric>
mth: how is the state of pandora?
<viric>
(I own a gp2x)
<viric>
and then there is that A-10 or so...
<viric>
what's the name of that small computer?
<kristianpaul>
nice the excap from dx have right chip !
<GNUtoo-desktop>
viric, I will need a sheevaplug like device soon
<GNUtoo-desktop>
is there a way to get a decent and reliable power supply?
<viric>
GNUtoo-desktop: the capacitors will explode, you just replace them
<viric>
GNUtoo-desktop: or.. it simply wants 5V. Half the sheevaplug is the powersupply, the other half is the computer
<GNUtoo-desktop>
yes but the big problem is that I can't do that if the box is in another country
<viric>
If you have a 5V 1.5A power supply it should work
<viric>
ah
<GNUtoo-desktop>
ok
<viric>
Then I would not get namely a sheevaplug
<viric>
they sell 'dreamplug' too.
<viric>
which has a power supply *apart*
<GNUtoo-desktop>
yes but in france hackable devices sell sheevaplugs
<viric>
not dreamplugs?
<GNUtoo-desktop>
no, I don't think so
<viric>
Maybe they improved the power supply silently... I don't know. Mine has around three years, maybe
<GNUtoo-desktop>
ok
<viric>
I know the guruplugs had a lot more troubles
<viric>
and the quickfix they did was to add a fan
<GNUtoo-desktop>
here's my use case
<viric>
GNUtoo-desktop: aren't you writing gnunet?
<GNUtoo-desktop>
I want a private email server
<viric>
I remember a 'gnutoo' from gnunet... maybe I'm confused though. :)
<GNUtoo-desktop>
I tried gnunet once
<viric>
GNUtoo-desktop: I use it as a private email server
<GNUtoo-desktop>
but it was some years ago
<GNUtoo-desktop>
so to get a mail server I need 2 box
<GNUtoo-desktop>
one here
<GNUtoo-desktop>
and one somewhere else for backup purpose, in case my box here goes down
<viric>
ok
<GNUtoo-desktop>
so the mail get delivered to the alternate MX record
<GNUtoo-desktop>
that would be my use of sheevaplug
<viric>
fine
<viric>
I tried to convince a friend to be a backup mx for me, but we still have not configured his side :)
<GNUtoo-desktop>
ok
<GNUtoo-desktop>
for me it would be my brother
<GNUtoo-desktop>
one day he's ok with it
<GNUtoo-desktop>
and the other day he's not
<viric>
MTA keep trying the sending three days...
<viric>
so I go on just expecting that I'll be able to bring the machine up in those three days ;)
<GNUtoo-desktop>
btw I need a domain name for it right?
<viric>
you need mx entries in some domain
<GNUtoo-desktop>
that's all?
<viric>
do you plan to run in from a dynamic ip?
<GNUtoo-desktop>
I tought that because of anti-spam rules I needed a more serious domain name than the one I have(dyndns)
<viric>
for a dynamic ip, the big problem is *sending mail* from it. But not receiving
<GNUtoo-desktop>
because I've setup my MTA(exim4) and my mail get refused
<viric>
GNUtoo-desktop: that's quite unrelated to the domain. It's related to your IP.
<viric>
I hire a sending service.
<viric>
(at dyndns btw)
<GNUtoo-desktop>
hmmm, but the website that blocked me proposed to unblock my domain name or something like taht
<GNUtoo-desktop>
*that
<viric>
I don't think so.
<GNUtoo-desktop>
I'm not sure they accept gnutoo.homelinux.org
<GNUtoo-desktop>
ok
<GNUtoo-desktop>
I'll ask to someone who has everything setup
<viric>
I have everything set up
<viric>
:)
<GNUtoo-desktop>
ok
<GNUtoo-desktop>
but you do not use your own MTA
<viric>
I use it, and it relies to dyndns; because dyndns owns an IP segment
<viric>
that's what the picky MTA receiving part checks
<GNUtoo-desktop>
<viric> I hire a sending service.
<GNUtoo-desktop>
ah ok
<viric>
I can afford a domain, but I can't afford an IP.
<GNUtoo-desktop>
I think I get it
<GNUtoo-desktop>
you forward to the dyndns MTA
<viric>
yes
<GNUtoo-desktop>
personally I want everything to run on box I trust to bypass my ISP MTA
<GNUtoo-desktop>
(for privacy reasons)
<viric>
GNUtoo-desktop: using your ISP MTA or not using it is irrelevant to privacy :)
<GNUtoo-desktop>
because it's not encrypted right?
<viric>
yes
<GNUtoo-desktop>
so all ISP monitor the content of SMTP transfers?
<viric>
they could
<viric>
having smtp transfers from your IP may be more suspicious than using the ISP mta ;)
<GNUtoo-desktop>
anyway I've a good associative mail but I prefer having my own mail because my associative mail is limited to 20M and they remove your account if you have more than 100M