<wpwrak>
adamw_: do you think you'll survive with this kind of setup for ~100 units ?
<adamw_>
for atusb 100 units?
<wpwrak>
yup
<wpwrak>
roh: (lasered antenna) seems that you have a lot of interesting research ahead of you :)
<adamw_>
this is "production fixture" I can make it.
<adamw_>
but my question won't be only that.
<wpwrak>
i'd send you a pair of these critters, ben cable and pogo pins
<adamw_>
1, find a free space to verify the transmission & receive performance is my concern.
<wpwrak>
i think they work reliably enough, just a bit unusual
<wpwrak>
yup, the testing process ... i now have a packet error rate test in the atrf utilities. so that should help.
<adamw_>
for a trial run, your this idea is quite enough. If need, I may make an real fixture to hook up<connection> without manually touching.
<adamw_>
hmm..sounds you already well-known.
<wpwrak>
for the free space, you probably don't need anything overly nice. probably just a large enough table and you're good for testing whether the board has been correctly assembled.
<wpwrak>
i.e., i would expect "bad" boards to fail in a very noticeable way, not just be a little worse. well, we can also just declare that "just a few dB" worse is okay for now :)
<adamw_>
so you need to have a near field test on your side first to show "PASS' or "FAIL' displayed on PC host program!
<adamw_>
then I survive! Or I dead. :-)
<wpwrak>
(near field) yes, that's the idea
<adamw_>
your report today is some equivalent 'far field' performance.
<wpwrak>
i did antenna sensitivity tests with about 3 m between sender and receiver in a largely empty room. not an anechoic chamber of course, but the best i have. then, for more interactive debugging, i just had things arranged in my lab. about 1 m between sender and receiver, lots of things laying around. and general behaviour in this "dirty" setup was similar to the cleaner one. so i think the near field isn't too bad
<adamw_>
so please write a good procedure to test 'near field' on said in 2 ~3 meter free space without large metal surrounding the circle.
<wpwrak>
and in addition, i did these "real life" far field tests, yes
<adamw_>
yeah...i saw your apartment layout! :-) It's real.
<wpwrak>
do you have a spectrum analyzer ? or if you don't, will you be able to borrow one ?
<adamw_>
so take some time to make some experimental data in 'near field' test.
<adamw_>
hmm...I don't have SA, I can try to borrow or go to some where to test.
<wpwrak>
for the field test, i was thinking of something like this: take one pc/laptop with a known to be good atusb. let this be the receiver. then put a known to be good DUT at whatever place you have (with laptop, for atusb, or ben, for atben)
<adamw_>
you should do a very clean laying around like 1 ~2 meters to get your data to be as threshold control to determine shown on PC side to indicate PASS or FAIL.
<adamw_>
:-)
<adamw_>
hmm...like that..
<wpwrak>
the send a test signal with maximum TX power and check the error pattern. adjust TX power down until errors start to appear.
<wpwrak>
then take this as a reference setting
<adamw_>
surely I am not sure your f/w needed to be calibrate to fix now...
<wpwrak>
(clear 1-2 m) yes, that would be good
<adamw_>
yeah..do a reference value for threshold setting , TX power, RSSI level.
<wpwrak>
if you can have a spectrum analyzer, it's easier to find any distortions. i.e., you'd see immediately if something is wrong
<adamw_>
you don't worry that SA, I can operate that. :)
<adamw_>
i just don't have it now.
<adamw_>
:-)
<wpwrak>
and you'd also see if you have changes caused by orientation or foreign objects. e.g., i've seen the spectrum change by about 10 dB just depending on where my body is.
<adamw_>
haha...
<adamw_>
I was always stayed in chamber before...
<adamw_>
well the better to find one here can support me. :-)
<wpwrak>
(sa) yeah, i know that you can that :) just convincing you that it's really useful to have for this
<adamw_>
not sure the new guy Wolfgang met can do or not?
<wpwrak>
unfortunately, i didn't think of telling wolfgang earlier about the spectrum analyzer. there's a company in germany that makes fairly inexpensive ones. he could have bought one and brought it.
<wpwrak>
naw, i mean for production testing. i suppose you'd want to do that locally, not send to germany etc.
<adamw_>
ha ..I didn't bought from OM.
<wpwrak>
well, joachim has a spectrum analyzer (at ccc), so that would be a plan B
<wpwrak>
the low-cost analyzer i was thinking of is only about EUR 400. but shipping would be expensive because it comes in a large box.
<adamw_>
hmm...you may need to think a schedule first...then to see if Wolfgang have idea for newest plan.
<adamw_>
i can get cheaper, just need to ask/check here.
<wpwrak>
(cheaper than EUR 400) ah, that's good. of course, you're at the source for low-cost stuff :)
<adamw_>
HF-4040?
<wpwrak>
yup
<wpwrak>
ah, sorry, EUR 500
<wpwrak>
they also have one for EUR 350, but i think the HF-4040 would be worth the difference
<adamw_>
hm..I won't have further action until Wolfgang tell me a real schedule. :-)
<adamw_>
hmm..i see.
<wpwrak>
(real schedule) also depends on your other activities, i guess
<wpwrak>
my plan is to make a few protoypes of the final design. then send a pair of atben/atusb to you as references, and one pair to roh as mechanical samples.
<wpwrak>
then you can play with the production test setup. ah, is it convenient for you to use two PCs for this or would you rather connect everything to the same PC ?
<adamw_>
to (same PC+ atusb) with (BEN + atben)
<wpwrak>
okay, then i need to add one option to my usb code to let you select which device to talk to if you have more than one connected to your pc. should have that already somewhere in my old openmoko stuff ...
<adamw_>
sounds good. :)
<wpwrak>
atben has no persistent state, so all it needs is the testing. atusb needs flashing. there are two steps: upload the boot loader with the pogo pin thing, then plug it into USB and download the rest of the firmware via DFU.
<wpwrak>
any later firmware upgrades can also be made with DFU
<wpwrak>
after DFU, it resets into the regular firmware, so you can then test without disconnecting
<adamw_>
ok,..just write proceduces in somewhere wiki..then I learn when we get schedule to start. the (real schedule) yes depends on other activities. :)
<adamw_>
right now I could not digest..:-) but it's real great to see your report today.
<wpwrak>
(document the process) will do that, yes. also make a little script around the commands or such
<wpwrak>
by the way, for the board cutting at the pcb maker, how do i make sure they're cutting the board precisely at the right place ? what i currently have is this:
<wpwrak>
(you need to view it with a monospaced font or it won't look right)
<adamw_>
digesting...
<wpwrak>
hmm, maybe i should make a PDF of it. just a sec
<adamw_>
hard to understand your 'stacking'...
<adamw_>
are u trying to say 'Don't cut antenna route?' while making pcb?
<adamw_>
in 'c2-closer.jpg'?
<wpwrak>
ah, not cutting the antenna is important too :) but no, i mean that the board cut needs to be very precise anyway, because it has to be shaped like a uSD card (8:10 card :)
<adamw_>
hm...when I get your sample and see your gerber, I will send to pcb maker to check. Two makers I dealed with now they have uSD card capability.
<adamw_>
so I may need to check the uSD's specification on pcb definition.
<adamw_>
i bet you already have such data, haven't it?
<wpwrak>
the uSD spec. yeah, measured it. with the help of Ornotermes' high resolution scan :)
<wpwrak>
the yellow thing is the PCB edge as drawn in kicad
<wpwrak>
there, i used a 5 mil line
<adamw_>
moment
<adamw_>
express picker is come...
<wpwrak>
now, my definition is that the center of that line shall be the real pcb edge. so the tool needs to be offset accordingly. the question is what the pcb makers expect. i.e., whether they can handle the line being technically zero-width, whether they need it to be offset by the line width, whether they also need a specific line width (to match the tool), etc.
<wpwrak>
updated the PDF. should be a bit more readable now
<qwebirc20929>
hi I need some help concerning unbricking my nanonote
<qwebirc20929>
wolfspraul may you please help me?
<qwebirc20929>
I used one euro coin in place of carbonized rubber that I have not
<wpwrak>
qwebirc20929: wow. isn't that huge ?
<qwebirc20929>
I also tried with a piece of coca cola litter
<qwebirc20929>
not too much
<qwebirc20929>
I thought It might connect the two pins
<wpwrak>
hmm, it has a rounded edge ... might be tricky
<wpwrak>
if you have tweezers, they may work better
<wpwrak>
or very small scissors
<qwebirc20929>
unfortunately I do not have here
<wpwrak>
or maybe take some aluminium foil, crumple it to give it some pointy peaks, and use that
<adamw_>
wpwrak, the dot line is you imagine the pcb edge (thus is the V-Cut area?)?
<adamw_>
wpwrak, i think i need git clone yours first. Otherwise I can't follow you.
<wpwrak>
the volume between the dotted lines is what the mill would remove, yes. you need a straight cut, not a V cut.
<wpwrak>
for the cutting, the CNC mill needs the path for the center of the tool. the question is how the pcb makers go from the board edge as drawn in kicad to this path.
<wpwrak>
depending on what they need, the data may need some prior conversion
<adamw_>
hmm..after I totally get your point, then you go further steps.
<wpwrak>
i have a tool to do that sort of math. doesn't output gerber yet, but that wouldn't be too difficult to add. (i used this tool with my mill, where i have the same type of transform to make)
<wpwrak>
sorry ;-))
<adamw_>
can you generate gerber then I check again?
<qwebirc20929>
I tried with a paper clip but the red led does not turn on for a while.. If I do not use the paper clip to connect the two pins the red led flashes for a while
<qwebirc20929>
I will go
<qwebirc20929>
to buy tweezers
<wpwrak>
this is the board outline. let me draw a top view
<adamw_>
yeah. draw a top view with atben pcb outline ..i still can't get your idea.
<adamw_>
there's a link at the bottom to check specification.
<wpwrak>
but the question is how the edge line gets interpreted by them. maybe they already interpret it in this way. maybe they offset the tool by half the line width. maybe they have a minimum line width. maybe they'll just adapt to whatever we tell them.
<adamw_>
if yours doesn't come with standard, you may need to modify..
<adamw_>
yes I see. I'll ask them that how they produce & cut it veritically.
<wpwrak>
the microSD mechanical specification is not open. only for SD assoc members :)
<wpwrak>
but i think my design is close enough :) handles exactly like a real uSD card.
<adamw_>
yes, could probably just alternatively use tool to change or they can handle.
<adamw_>
really?...not open? hmm...interesting.
<wpwrak>
well, the real card is a bit thinner - 0.7 vs. the 0.8 mm of my PCB. but that's something i can't control. besides, the 0.8 mm may actually make it wobble a bit less
<wpwrak>
so with the edge, there are basically three questions:
<wpwrak>
1) can they do this accurately enough at all ? we shouldn't be the only ones with narrow tolerances, but not every pcb shop may be able to handle this.
<wpwrak>
2) how do they handle the tool position with respect to that PCB Edge line. in particular, can they align it such that it cuts along the edge ?
<qwebirc20929>
I found the tweezers
<wpwrak>
3) are there any limitations for the PCB Edge line ? e.g., do they require a specific width ? (this would be the case if they treat it directly as a toolpath without any conversion)
<wpwrak>
qwebirc20929: whee ! should be easier now. just make sure you disconnect usb for ~30 seconds before trying
<qwebirc20929>
so the sequence3 is the following ..
<qwebirc20929>
step one disconnect usb cable from both side
<qwebirc20929>
step two connect the two pins with the tweezers
<qwebirc20929>
step tree connect the nn to notebook while the pin are connected by tweezers
<qwebirc20929>
the pins sorry
<wpwrak>
qwebirc20929: yup, sounds good
<qwebirc20929>
then have a look at watch lsusb terminal
<wpwrak>
qwebirc20929: it's probably difficult to keep the pins shorted while connecting the usb cable to the ben. so i'd add
<qwebirc20929>
there is no need to press the on button
<wpwrak>
step 1.5: connect the USB cable on the Ben side
<adamw_>
wpwrak, I'll let you know after I know.
<wpwrak>
that way, when you connect USB, you don't have mechanical forces working against the tweezers
<wpwrak>
adamw_: thanks !
<wpwrak>
qwebirc20929: the on button is not needed
<wpwrak>
qwebirc20929: important: the time while the ben has no power should be ~30 seconds. otherwise, it may not properly discharge and thus not reset.
<qwebirc20929>
please add it in the wiki the issue of 30 seconds
<wpwrak>
congratulations !
<qwebirc20929>
beacuse it sounds crucial
<wolfspraul>
qwebirc20929: I see wpwrak already helped you unbrick your Ben. wonderful!
<wpwrak>
wolfspraul: seems that time time-to-drain is missing (the 30 secs) ?
<qwebirc20929>
it is not mentioned
<wpwrak>
s/time //
<qwebirc20929>
on the wiki
<wolfspraul>
30 seconds sounds like a lot, I doubt more than 10 is ever needed
<qwebirc20929>
so users could become frustrated
<wpwrak>
would be good to add. it's a bit longer than most people would intuitively assume
<wolfspraul>
qwebirc20929: you don't have the carbonized rubber button? (sorry long backlog haven't read it all yet)
<wpwrak>
wolfspraul: i think it's more than 10 sec. something around 20. 30 is with a bit of safety margin.
<qwebirc20929>
no sir
<qwebirc20929>
I used the tweezers
<wolfspraul>
why not? it's included with every Ben. comes in a small plastic bag.
<wolfspraul>
the risk of tweezers is that you apply to much force and scratch the board, or scratch/rip off the pads
<qwebirc20929>
It was not in the bug
<wolfspraul>
but you made it already :-)
<qwebirc20929>
bag sorry
<wolfspraul>
strange
<wpwrak>
one item for the ya wishlist: connect that usb boot to a real button ;-)
<wolfspraul>
did you see a small plastic bag?
<qwebirc20929>
yes I did
<wpwrak>
(that is, if we still need it then)
<wolfspraul>
wpwrak: we tried, but it was surprisingly hard because of some timing issue.
<qwebirc20929>
but there was not
<wolfspraul>
the bricking problem should mostly go away if our software were more robust
<wpwrak>
wolfspraul: timing issues with a button ? :)
<wolfspraul>
I blame usbboot :-)
<wolfspraul>
wpwrak: it's a long story, seriously both Adam and Carlos tried but failed to make it work...
<qwebirc20929>
someone forgot or maybe it dropped while I was exploring it for the first time
<wpwrak>
wolfspraul: then people will write new, less robust software ;-)
<qwebirc20929>
now I would install this openwrt-xburst-qi_lb60-u-boot.2010.07.07.bin
<qwebirc20929>
beacuse I would like to boot 4gb with debian
<wpwrak>
wolfspraul: hmm. you'd have to connect the button to some gpios .. if there's weird stuff on these at reset time, that could cause trouble. but that's the only issue i could think of.
<qwebirc20929>
on the internet is written that is better for 4gb microsd card isn't rue?
<wolfspraul>
and why did you try to reflash a kernel into page 0, overwriting u-boot?
<wolfspraul>
wherever you get your instructions from, it's wrong :-)
<kyak>
most people who have bricking problems with Ben were giving a try to debian.. maybe, in fact, there are wrong instructions for flashin debian into Ben spreaded around the Internet?
<qwebirc20929>
so please inform the author of that web page
<qwebirc20929>
or I will tell him when I will have time
<wpwrak>
kyak: on last count, i think it was 100% debian ;-)
<kyak>
wpwrak: yeah, i'm just trying to find possible excuses for debian :)
<wpwrak>
kyak: well, there was the nerase bug. there probably are still many broken usbboots in circulation
<wolfspraul>
xiangfu: that's your page
<qwebirc20929>
okay may I run reflash.sh script now?
<wolfspraul>
you can try. you already went through the worst possible experience, it cannot get worse!
<wolfspraul>
you are hardened now - no more fear.
<wolfspraul>
worst case you overwrite u-boot again and have to short the pins again :-)
<wpwrak>
with the ben, the only thing to fear is fear itself :)
<wolfspraul>
xiangfu: you should correct or delete the page with wrong information (I cannot load it right now so I don't know what it says...)
<wolfspraul>
or just put a link to better instructions there
<wolfspraul>
if you want to be perfect you can leave the old page as is, but add a clearly visible note at the top that the information is outdated, with a link to a better page
<xiangfu>
oh.. but here are a lot of old page in that blog. I will add the wiki link to that page.
<xiangfu>
wolfspraul: yes.
<xiangfu>
wolfspraul: that is better :)
<wolfspraul>
sure no rush, just in principle.
<wolfspraul>
misleading information is much worse than no information at all.
<wolfspraul>
of course we are all dealing with progress, and there is tons of outdated stuff everywhere
<xiangfu>
about build progress. now there are only two packages not compile in build hosts: php5 and plplot
<xiangfu>
also I update my build script file. using make "make package/symbiclink" not "./scripts/feeds ..." anymore
<xiangfu>
I need go. see you guys later.
<qwebirc20929>
I would understand how to manage openwrti graphic interface
<qwebirc20929>
wolfsspraul reflash.sh has finished its job
<qwebirc20929>
What is the correct sequence now?
<qwebirc20929>
unplug usb cable
<qwebirc20929>
reinsert battery
<qwebirc20929>
and then press on button and wait for 20s
<qwebirc20929>
Am I right?
<wolfspraul>
sure sounds good
<wolfspraul>
don't hesitate this is a consumer quality product, you can play with it a little, it will not break
<wolfspraul>
if anything then the software is not consumer quality yet...
<wolfspraul>
qwebirc20929: is it booting?
<qwebirc20929>
just a second please
<qwebirc20929>
it is sir
<wolfspraul>
good
<wolfspraul>
that's how it should be
<wolfspraul>
I don't know what your plans are, but very roughly I would suggest you play with this a litle, go through some apps.
<wolfspraul>
if you like adventures, you can then also try Jlime
<wolfspraul>
but get a bit more familiar with the device first
<qwebirc20929>
I I have run ebook but the nn is freezing
<qwebirc20929>
I can not use ctrl alt f1 etc to switch to terminal
<qwebirc20929>
I sounds like nn stopped itself
<qwebirc20929>
now the screen is off
<qwebirc20929>
and it does not turn on anymore
<qwebirc20929>
perhaps is the battery
<wolfspraul>
one by one
<wolfspraul>
which ebook app did you run?
<wolfspraul>
some apps have a slow startup time on the first launch after reflashing
<wolfspraul>
up to 60 seconds (is a bug, should be improved)
<wolfspraul>
like stardict for example
<qwebirc20929>
now is turning on again .. with usb cable
<wolfspraul>
you can try ctrl-alt-f2
<qwebirc20929>
abook
<wolfspraul>
the device does not properly shut itself down when the battery is low, I believe
<wolfspraul>
in that case it will just drain the battery, until the screen turns white and the devices goes off
<wolfspraul>
a bit rude :-)
<qwebirc20929>
yes I can
<qwebirc20929>
I can see busybox
<qwebirc20929>
now I would run xiafu debian
<qwebirc20929>
I will write halt
<qwebirc20929>
and then on + M together Am I right?
<wolfspraul>
that boots from the flash card, right?
<qwebirc20929>
hope so
<wolfspraul>
I believe it does.
<wolfspraul>
Debian will be very slow, in my opinion it's unusable on the Ben.
<wolfspraul>
just for demo purposes :-)
<qwebirc20929>
I will set up 1gb of swap .. that should help
<kyak>
help making it even more slower :)
<qwebirc20929>
it did not boot
<qwebirc20929>
now I have openwrti
<qwebirc20929>
I have another question
<qwebirc20929>
I order on the web some nokia batteries 3.7 V 850 mmAh
<qwebirc20929>
but I am not sure if they will fit on nn
<qwebirc20929>
Do you have any hints on battery for nn ..?
<qwebirc20929>
sorry any hint
<kyak>
there is a compatiblity list on wiki
<qwebirc20929>
the second thing is about building a nn cluster using an usb hub... Is it possible?
<kyak>
if i said "it is possible", would you know how to do it? :)
<qwebirc20929>
I know how to do it with a regularar 386 amd64 mixed cluster using a classical ethernet hub
<qwebirc20929>
but here the archicterure is different
<kyak>
since Ben has Ethernet over USB capability i guess yes, it is possible
<mth>
iirc the NN is a USB device, not a USB host
<mth>
you'd need one host device on such a network, I think
<qwebirc20929>
sure
<kyak>
yeah, not a host
<qwebirc20929>
yes
<qwebirc20929>
sure
<qwebirc20929>
I would be challenging to do it
<qwebirc20929>
beacuse the energy consumption is lower
<qwebirc20929>
and the principles are the sames of bigger 386 amd64 brothers
<qwebirc20929>
because sorry
<kyak>
very interesting to know the results, don't hesitate to post them
<mth>
you could use a netbook, a usb hub and a bunch of NanoNotes
<qwebirc20929>
yes sir that is one possible idea
<qwebirc20929>
I would install gnu OCTAVE
<qwebirc20929>
on debian is still possible
<qwebirc20929>
and easy to do
<qwebirc20929>
all the packages are ready
<qwebirc20929>
for xburst arch mpisel tec..
<qwebirc20929>
ok
<qwebirc20929>
Now I have to go Thanks to all for your patience and help
<zrafa>
bah wolf is not more here..
<wpwrak>
maybe having dinner
<zrafa>
which applications are the stars for nn?
<Jay7>
just network outage (ping timeout)
<Jay7>
zrafa: prboom! :)
<zrafa>
wpwrak: I am asking because I always read that Debian is slow and unusable.. But, on the other hand I see that official openwrt qi version is full of applications for console
<zrafa>
wpwrak: so I guess that those applications would work okey using Debian
<zrafa>
wpwrak: and also, those would have full features and no less
<Jay7>
and that console organizer, noticed in ML.. can't remember it's name
<Jay7>
some 3 letters
<Jay7>
something like 'nhm'
<zrafa>
Jay7: prboom is light.. it should work okey in Debian on fb
<zrafa>
I need to test Debian soon
<zrafa>
bc, top, vim, etc.. none of them should be slow in Debian. I wonder why is unusuable
<wpwrak>
zrafa: (slow debian) maybe an unoptimized boot process or such. can't quite imagine that the "normal" apps would be much slower than on other distros
<wpwrak>
zrafa: maybe fatter, with man pages and stuff
<zrafa>
wpwrak: yeah, but that does not do Debian slower (more files)
<zrafa>
wpwrak: on the other hand it seems that if I test Debian my nn will have a broken bootloader after installation :P
<wpwrak>
zrafa: well, you have idbg. with it, the usb boot is very very easy ;-)
<wolfspraul>
wpwrak: if we wanted to embed atben into a future product, how hard is it in terms of antenna tuning, or reusing what we have currently in general?
<wolfspraul>
can we reuse the layout? or totally new layout needed?
<roh>
uff
<wolfspraul>
can we cut & paste the antenna? how about the PCB specs?
<roh>
drilled all the shielding sheets
<wpwrak>
totally new layout. you also have more layers, so you make everything denser. yes, you can reuse the same antenna.
<wpwrak>
after that, you need to measure ;-)
<wolfspraul>
roh: congrats! how long did it take you?
<wpwrak>
the "core parts" seem to be pretty stable. even if you mess with the layout a lot, nothing really changes.
<roh>
wolfspraul: didnt stop it
<roh>
i wrote some gcode and used the mill
<roh>
it fits just like that. 3mm more and i couldnt use the mill ;)
<roh>
made some pics
<wolfspraul>
nice
<wolfspraul>
gcode! :-) I vaguely remember this from somewhere...
<wolfspraul>
for the sake of it, commit it along with the dxf :-)
<wpwrak>
roh: do you write gcode or just generate it ?
<roh>
this time i just wrote it down
<roh>
easy enough for 4 holes
<wpwrak>
ah okay. no big deal then :)
<roh>
it takes more time to manually remove the splinters
<roh>
'entgraten'
<roh>
and i dont have stickers to isolate yet. i got a dxf for it.. will find out if it makes more sense to lasercut or use a cutting plotter for that
<roh>
this shielding thing eats much more time than initially thought
<wpwrak>
(splinters) that would be burr. the process is called deburring. and yeah, do i hate it, too :)
<wpwrak>
(isolate) why not do that before cutting so you cut both at the same time ? may also reduce the burr
<roh>
nah. i need clean metal around the holes
<roh>
thats where the sheet connects to the shielding of the board via the metal screws and distance pieces
<wpwrak>
ah, i see
<wpwrak>
roh: btw, have atmel released their debugWIRE spec by now ? or has it been at least reverse engineered ?
<steve|m>
wpwrak: neither nor afaik
<wpwrak>
steve|m: :-( the large number of signals needed for in-circuit programming is a bit annoying ...
<steve|m>
wpwrak: is programming via dW even possible?
<steve|m>
wpwrak: and then again, how do you enable the debug wire fuse?
<roh>
wpwrak: dunno.. never used that
<wpwrak>
steve|m: hmm. good question :)
<wpwrak>
roh: you said that the AVR USB chips are well supported by open software. so there's a basic USB stack ? what I need is just control transfers. an existing DFU implementation would be nice. or else, i could port the one i did for the f32x.
<wpwrak>
sounds quite decent. has a lot more than i need :)
<roh>
its the atmel flavour of dfu i think
<wpwrak>
i like this: "my first foray into the world of USB" [...] "LUFA is being used all over the world, in many applications - both free and commercial."
<roh>
dunno whats the difference
<wpwrak>
as i understand it, LUFA is the whole stack. and there's DFU in there too.
<roh>
ack
<roh>
there are just differen flavours of dfu it seems
<wpwrak>
ah, that would be stefan's problem, no ? ;-)))
<roh>
hrhr
<wpwrak>
well, if atmel did anything dirty, we can also undo it. the joy of having sources :)
<wpwrak>
(re-sealing may be a different story, though :)
<bartbes>
for the horny ones among you: there's a better use for silicone
<roh>
heh
<wpwrak>
bartbes: let's just say that there are many reasons for men to appreciate silicone ;-)
<steve|m>
dipping in hot-glue should work fine as well :)
<wpwrak>
roh: (the pics are the same as in the mail i just sent to the list)
<wpwrak>
steve|m: ah, that would be another option. haven't played with that stuff yet.
<bartbes>
wpwrak: and those are swhich modules?
<wpwrak>
bartbes: oh, various failed experiments. a few idbgs that went sour (the small ones), then one ftdi-bases c2 programmer that never worked, probably due to weirdness in the ftdi chip itself, and then one atusb that may have suffered braindamage due to repeated thermal stress.
<bartbes>
I'm really looking forward to atben and atusb
<wpwrak>
good ! :)
<wpwrak>
steve|m: how hot does hot glue normally have to get before reaching minimum viscosity ?
<wpwrak>
ah, i think i found it. polyamides, 200 C. not too bad.
<wpwrak>
a bit more expensive than silicone (again, bathroom sealants) and also a bit harder to use (it thicker and thus needs more pressure)
<bartbes>
wpwrak: so what kind of data are you transmitting in your tests?
<wpwrak>
roh: of course, with the tiny quantities needed here, the material cost is pretty negligible. you can probably do hundreds of pieces with just one catridge
<wpwrak>
bartbes: in the ping test, just a locally generated sequence number, the last received sequence number, the crc, and a lot of uninitialized bytes for padding
<wpwrak>
bartbes: in the PER test, it's memset(packet, seq, 127); with "seq" a 8 bit counter. again, the sequence number
<bartbes>
so no protocol at all, really
<wpwrak>
bartbes: note: no CRC in the PER test. i then do a statistical analysis of the bytes to find which number i really have there. then i scan the packet for bad symbols (IEEE 802.15.4 used four-bit symbols that get send as one 16-bit "chip")
<wpwrak>
just my protocol ;-) the bare essentials. no addresses or such. just length, my stuff, and maybe a CRC
<roh>
wpwrak: depends
<wpwrak>
anything else would only get in the way
<roh>
some molding stuff goes bad when opened and needs to be used quite fast
<bartbes>
wpwrak: so you can't even guarantee we can browse teh internetz wirelessly?
<wpwrak>
(symbols/chips) that is, for the band we use here. 802.15.4 has a gazillion other modulation techniques, some quite different.
<wpwrak>
bartbes: nope, i can't guarantee it. if lightning strikes me today and everyone else who could connect the dots in the kernel tomorrow, then you can't :)
<bartbes>
wouldn't it be worth checking whether using 6lowpan actually yields results you can live with before going into mass production?
<wpwrak>
roh: sure. but you can just squeeze a little bit of the sealant out of the cartridge. then it lasts a while. it will eventually also cure inside the cartridge, but you have days if not weeks before that becomes a problem.
<bartbes>
(i.e. do a limited number first, send those out to the devs, then do a full production later?)
<wpwrak>
bartbes: i don't really expect any problems there.
<roh>
wpwrak: i meant the pur primers
<roh>
the 2 component stuff
<wpwrak>
roh: ah, but now you're talking about sophisticated things ;-) anyway, there, the clock usually doesn't start ticking until you mix them. so just don't mix the whole cans at once :)
<wpwrak>
bartbes: the planned number will be relatively low anyway. 100 pieces. doesn't really make sense to go below that, due to high fixed costs.
<wpwrak>
bartbes: this will also be without certification and such. so sold as a development/experimental kit, not as an end-customer solution
<wpwrak>
bartbes: for certification, we'll need a real RF lab to look at things. this will also happen eventually, but all in due time
<wpwrak>
bartbes: the idea is to get the design we have out now, to get things rolling. beat the software into shape, gather usage experience, maybe collect some ideas for improvements, and all that.
<bartbes>
aww
<bartbes>
I want one
<wpwrak>
bartbes: no point in spending a decade on finding the perfect recipe for brioche while the masses are starving (with apologies to marie antoinette ;-)
<bartbes>
so I can use irc while on the can
<wpwrak>
yeah. that's roughly my goal, too ;-)
<bartbes>
coolest would be if I can do.. like computer maintenance jobs around the house
<bartbes>
and yet always have access to a set of linux tools
<bartbes>
so I can at least ping and stuff ;)
<bartbes>
(I'd probably ssh into my own comp, and do stuff from there)
<bartbes>
I could get fat, not walking up and down all those stairs
<wpwrak>
well, it's a bit like wifi. not quite as fast, and doesn't go quite as far, but you shuold basically be able to do all the same things.
<bartbes>
wee
<wpwrak>
one little difficulty will be that 6LoWPAN is IPv6. so if you're not IPv6-capable, you need to make this transition, too. haven't looked into the details yet. even if the home network is IPv4-only, there ought to be some NAT-like translation mechansisms to run on the PC acting as gateway.
<bartbes>
I'd be so lazy I'd put a laptop somewhere close with the usb device if I have to
<bartbes>
don't know about that
<bartbes>
I know all linuxes are ipv6 enabled
<bartbes>
well, modern ones
<bartbes>
and windows has been for quite a while
<bartbes>
and though my router doesn't support it
<bartbes>
it's simply bypassed
<bartbes>
I occasionally use ipv6 connectivity on LAN already
<bartbes>
but it's kind of useless
<bartbes>
since I know the ipv4 addresses
<bartbes>
but the ipv6 ones are.. quite a bit harder
<bartbes>
;)
<wpwrak>
yeah, ipv6 addresses are good memory training ;)
<roh>
there are openwrt routers which can be hacked to also do usb for about 18E
<wpwrak>
roh: some wpan-wifi gateway would be nice to have indeed
<roh>
if there is source somebody can package it for openwrt -> tadaa.wav
<wpwrak>
roh: you just have to make sure wifi is on a channel very far away :)
<larsc>
there are even routes with usb device these days
<wpwrak>
usbnet, i suppose ?
<kristianpaul>
[silicone] dint look soo good
<kristianpaul>
may be a second layer will improve shape
<wpwrak>
the silicone just follows the shape of the components. and yes, i think a second layer may make it nicer.
<wpwrak>
grr. seems to be hard to find wireless routers that are cheap, have usb host without soldering, that are supported by openwrt, and that you can get in argentina. there's a gazillion of tp-link here, all dirt-cheap, but they're either not supported or have no usb host. grmbl.
<kristianpaul>
fonera?
<kristianpaul>
ah argentina, thats the barrier
<rjeffries>
werner: I like the silicon encapsulation idea. simple, cheap, nerdy
<wpwrak>
rjeffries: yeah, it's lowly growing on me. particularly after i discovered that it can be cut open quite cleanly
<wpwrak>
s/lowly/slowly/
<kristianpaul>
hey, you should ask to the buenos aire libre people, they may have or be aware of usb-host capable wifi router
<rjeffries>
werner: Dockstar + atusb could be interesting. 3 full size USBs + 1 mini USB, a 1GHz ARM processor, 128MG RAM, Ethernet plus a couple of LEDs.
<rjeffries>
Dockstar is low power ~4 watts. a lot of people are hacking it.
<wpwrak>
a bit pricy in comparison. but a tidy package.
<wpwrak>
ah wait ... doesn't have wlan
<wpwrak>
but yes, for a wpan-to-ether gateway, why not. but it's probably easier to use a wlan router. that has ethernet too, plus all the rest.
<kristianpaul>
plus openwrt :-)
<roh>
_
<lekernel>
rapidsmith looks pretty good (except that it's written in Java)... it's running the xilinx tools with various options to get complete maps of the actual fpga chips :)
<lekernel>
they're definitely on the right track
<wpwrak>
roh: what's your opinion on a simple silicone coat ? technology transfer may be a bit difficult in this case, but then, it as low-tech as we can get ;-)
<wpwrak>
s/it/it's/
<roh>
dunno yet...i like something which gives hard surfaces
<roh>
i dont like it when pcbs get mechanical stress
<wpwrak>
they'll get plenty of that anyway :-( at least the atben board. the only thing holding it is a bit of PCB going into the 8:10 card slot. and it has a pretty long lever to give it a good push every now and then.
<wpwrak>
atusb is safer, though. it's just a slightly oversized usb stick
<roh>
wpwrak: to be fair.. i'd wish to have mold-able abs+pc mix like used as in regular cases
<wpwrak>
hmm, sounds like a lot of work
<wpwrak>
roh: at least you have the heavy machinery for rapidly cutting metal molds. with my little mill, i prefer wood. beats crunching aluminium for days ;-)
<roh>
heh. yes.
<roh>
can just take a 16mm diameter cutter and an aluminium block and go nuts
<roh>
still takes time but eats through nicely
<wpwrak>
meanwhile, i slowly grind with my 35 mil endmill at 0.2 mm/s or so, maybe doing 0.5 mm (Z) at a time ...
<wpwrak>
when my disk died a while ago, i actually lost my list of feed rates. so i'll have to determine those again :-(
<wpwrak>
(access points) the linksys/cisco WRT160N is also available here. it's listed as work in progress at openwrt
<rjeffries>
back to Dockstar: OpenWrt is running on it if I did not make that clear. I happen to use Debian on one, Plugapps on the other
<wpwrak>
ah, now i found an 610N. expensive, though, USD 180 (local price)
<wpwrak>
nice way to gather e-mail addresses ;-)
<wpwrak>
rjeffries: yes, but, say, a TP-Link TL-MR3420 or MR1043ND would have quite similar features, plus WLAN. similar price.
<wpwrak>
rjeffries: and such wireless routers can be found worldwide, without expensive shipping. even at remote places like argentina ;-)
<rjeffries>
wpwrak that's cool. I got my Dockstars for $20 each so that was fun. agree that the wifi router as an extender or base station would be A Good Thing.
<rjeffries>
<wpwrak> nice way to gather e-mail addresses ;-) //what are you referring to?
<roh>
wpwrak: we can send you some cheap hw as carepackets ;)
<roh>
removing the psu before
<wpwrak>
rjeffries: (gather addresses) that mailing list thing. no archive, so the only way to find out what's there is by subscribing. nice how people learn from facebook :)
<wpwrak>
roh: actually some HP paper would be nice :) i'm approaching my last sheet. and i'm not looking forward to re-evaluating my toner transfer process ...
<roh>
hp-paper? got a link?
<wpwrak>
and it seems that HP have stopped selling their paper in argentina :-(
<wpwrak>
lemme find the name ...
<wpwrak>
C6039A
<wpwrak>
also available in a larger package as C6979A
<roh>
wpwrak: uh.. thats letter... cant buy that here
<wpwrak>
yeah. dunno if they have the same paper in A4 as well. i guess they should.
<zrafa>
wpwrak: we have tried another paper for ledtoy toner transfer which worked very well
<roh>
wpwrak: doesnt amazon.com deliver for you? ;)
<wpwrak>
zrafa: ah, which one ?
<zrafa>
wpwrak: let me check..
<wpwrak>
roh: amazon should work, yes. do they sell paper now ? checking ...
<zrafa>
wpwrak: glossy inkjet photo paper.. Our has the title and description in spanish
<zrafa>
wpwrak: but the front is the same
<wpwrak>
zrafa: ah, looks interesting. where did you find it ?
<zrafa>
wpwrak: no remember, but I remember I saw it on several book shops or similar shops which sell stuff for school
<zrafa>
wpwrak: we have around 15 sheets yet, so if you want to try we can give you those
<roh>
i wonder if it would be cheaper to send it from somewhere else in latin america
<wpwrak>
roh: probably not. then you pay shipping to that place, their outrageous taxes, plus shipping to argentina. and often enough, the latter may even go via the US ...
<wpwrak>
zrafa: i may come back to that offer :) thanks !
<kristianpaul>
wpwrak: i do tonner transfer re-using sheets from a local magazine :-)
<wpwrak>
kristianpaul: playboy colombia ? ;-)
<kristianpaul>
lol, not that one
<kristianpaul>
well i used to, long time since i dint :(
<kristianpaul>
I need explore dual layer thin pcb some day :-)
<kristianpaul>
But so far i dint had luck sourcing 0.8~ mm PCB
<zrafa>
wpwrak: if no playboy colombia then maybe playboy venezuela
<Jay7>
hm.. good idea to re-use playboy to transfer toner :)
<Jay7>
it even may transfer some parts of images by accident :)
<roh>
wpwrak: how do you solder the parts on the atben/usb?
<roh>
hot air soldering station?
<roh>
or pizzaoven?
<roh>
to be fair.. i like the idea of atusb having a mini-usb-b socket instead of a usb-A plug directly
<roh>
hm. but one could also use a usb A male - female cable for that.
<roh>
i guess usb-A is more common
<Jay7>
which connector is used in mobile phones now?
<Jay7>
mini-B?
<wpwrak>
kristianpaul: (0.8 mm) digi-key PC44-NDÂ Â PCB COPPER CLAD 3 X 4.5" 2 SIDE
<wpwrak>
roh: such cables do exist. i got some from digi-key :)
<wpwrak>
Jay7: mini and micro
<wpwrak>
roh: (solder) just iron. except for the crystal. there i use hot air.
<Jay7>
just use some then :)
<wpwrak>
roh: i get a lot of air pressure with hot air, so this blows away the solder. not sure if my station really works as it should. but it's good for removing excess solder ;-)
<kristianpaul>
wpwrak: i know, but i dont like digi-key as they dont send stuff by cheap postal mail...
<wpwrak>
roh: my toaster oven has trouble maintaining a good temperature. so i only use it on days when i feel particularly foolish :)
<wpwrak>
kristianpaul: are you sure they don't offer US parcel service ?
<kristianpaul>
wpwrak: let me re-check
<wpwrak>
kristianpaul: they do to argentina. of course, they have a great fedex rate too, so i now always use fedex
<kristianpaul>
wpwrak: do postal but is EMS
<roh>
hmmh
<kristianpaul>
i need first class postal mail
<wpwrak>
kristianpaul: so, no advantage ?
<kristianpaul>
32.5 is not on my budget
<kristianpaul>
USD^
<kristianpaul>
wpwrak: no :(
<wpwrak>
USD 32.5 doesn't sound too bad ...
<kristianpaul>
for me they do :-)
<wpwrak>
buy more one item at a time :)
<wpwrak>
more than
<kristianpaul>
yeah that probably a solution
<kristianpaul>
s/that/thats
<rjeffries>
werner said: <wpwrak> rjeffries: (gather addresses) that mailing list thing. no archive, so the only way to find out what's there is by subscribing. nice how people learn from facebook :)
<rjeffries>
wpwrak: that is a fait criticism. Tintletter.com is a new service, run by a one man band. I have already requested a way to allow now subscribers to peek.
<steve|m>
wpwrak: The LP2985 is shut off by driving the ON/OFF input low, and turned on by pulling it high.
<wpwrak>
ooh .. there ! sorry :) everybody else has Vin in the upper left corner ;-)
<roh>
wpwrak: the lp2951 has also an on/off, is avail as fixed voltage version and can give you a overcurrent-error-output statis
<rjeffries>
wprak have you considered adding a multicolor LED so atben and atusb indicate receive/transmit?
<wpwrak>
lp2985 is even pin-compatible with the one from analog. nice :)
<wpwrak>
roh: lemme see ..
<roh>
depends if thats interresting to you
<wpwrak>
pin-compatible with something else is always nice :)
<wpwrak>
but the 2951 looks good indeed
<roh>
65cent or so
<wpwrak>
roh: perfect. thanks a lot !
<wpwrak>
rjeffries: naw. no i/o on atben. atusb at least gets a led, but just unicolor. it's currently only used for indicating boot state.
<wpwrak>
rjeffries: (boot state) when the mcu resets, it listens for about 1 s on USB for anyone trying to speak DFU to it. after that, it jumps to the application. while listening for DFU, the LED is on.
<wpwrak>
heh, i still have a few LP8345. my, is that old junk ...
<rjeffries>
wpeak understood. on atusb I like the boot state led, good stuff. adding led(s) so transmission and recive are indicated would be awesome, cost impact is small, usefulness is high IMO
<wpwrak>
rjeffries: there are no i/o lines available for this on atben. also, the board is rather crowded as it is. i'm fighting for every tenth of a millimeter there :)
<wpwrak>
i don't even have enough i/os to drive reset. fortunately, power-on reset works very well
<wpwrak>
besides, you have a LED right on the other side of the ben. and 76800 colorful pixels all within easy reach ;-)
<rjeffries>
wprak understood re atben. hope you will consider enhancing atUSB slightly. ;)
<wpwrak>
rjeffries: naw, doesn't really make sense. if you really want an activity led, it's already there. tx and rx happen in pairs anyway, and you won't be able to see a difference of a few milliseconds.
<wpwrak>
steve|m: i tried.  but it didn't go well. the two boards i made both acted mysteriously
<wpwrak>
steve|m: perhaps too much thermal stress by putting the board on and off the iron
<rjeffries>
wpwrak said "tx and rx happen in pairs" Agree, they are supposed to happen in pairs, but in a range challenged situation or noise rf environment one could easilly transmit but not get anything bacl. That is interesting. ;)
<wpwrak>
well, you could flash the led when an ack is missing. (ieee 802.15.4 always acks packets, much like wlan does)
<wpwrak>
mmh, fedex seem to be having a bit of weather trouble. can't imagine why. been nice and warm all day around here :)
<kristianpaul>
wpwrak: From you experience with CNC machines, how good is share the "power line" in wich it is connected, with let said other devices like a Milkymist One board and a scope :-)
<wpwrak>
not a problem at all
<kristianpaul>
I'm running out space here now, but i started an automatic loong print job,
<wpwrak>
well, unless you have some real monster at home :)
<kristianpaul>
you alredy know it, 3 steppers 1 DC motor, and some heaters
<kristianpaul>
I just wanted to be sure from experiene as i cant measure armonics in AC :/
<wpwrak>
i keep my air conditioners away from the rest, but i don't worry about my mill. my mill draws as much power as a netbook.
<kristianpaul>
oh sure
<wpwrak>
you can check the power supply. it's probably quite benign.
<kristianpaul>
Is a standard ATX 400W power suply
<wpwrak>
wrong ! now that you have a scope, you can ! ;-)))
<wpwrak>
(atx 400 W) greedy beast :)
<kristianpaul>
Hey, yes !
<kristianpaul>
but is not my priority now
<wpwrak>
my mill is 19 V, 2 A
<kristianpaul>
with air conditioners you mean ventilador?
<wpwrak>
the thingies that dry the air, cool it, or heat it. and also move it around, yes.
<kristianpaul>
hmm i'll move it right now it shares main with cnc...
<kristianpaul>
Was a whole 30°C day here today :S
<wpwrak>
the main risk are power consumption peaks, i.e., when the air conditioner switches itself on. if you have a fan that just runs the whole time, then that's okay.
<wpwrak>
also, noise on the AC line should get filtered anyway. but of course, if something draws enough current to make the voltage drop, then some devices may get upset
<wpwrak>
e.g., if i switch on enough air conditioners, my desktop pc resets from time to time. that is, it did before i added the usp. works like a charm. so the one i had before was just a bad device.
<kristianpaul>
well, my PC have UPS as well :-)
<kristianpaul>
just CNC and scope and m1 are out of the backup due load capacity
<wpwrak>
if your cnc does very long-running jobs, then an UPS may be useful (as long as the PC controlling it is also supplied)
<kristianpaul>
btw i now can print continuoslly with my 3d printer !, i just added a Conveyor system, and fixed it so it works better now for ie 7 hrs of printing :-)
<kristianpaul>
All run from SD so PC is not processing nothing besides the start command
<wpwrak>
the conveyor ought to be nice :) i'm glad i don't have anything like this. otherwise, i wouldn't rest before i had hatched some crazy project that makes the conveyor work for at least a week non-stop :)
<kristianpaul>
he, i a week i will run of plastic
<wpwrak>
f(2.95, 3.25) = 3.1
<wpwrak>
f(3.45, 3.75) = 3.5
<wpwrak>
what is f() ?
<wpwrak>
(this is the magic qfn pad to land pattern function nxp use)
<wpwrak>
hmm, the tendency is that the number of different and mutually contradictory design parameters for QFN packages increases linearly with the number of sources consulted.
<wpwrak>
at least i'm now almost entirely convinced that i'll trust only half of adam's rule of thumb :)