<wpwrak>
whitequark: you measured that you don't have shorts ?
<whitequark>
traces are 10mil. the soldering looks perfect: I may even place a light behind the board and see how traces and chip pads align to each other
<whitequark>
sure
<whitequark>
checked all the three identical boards a lot of times
<whitequark>
for shorts, openings, everything
<wpwrak>
whitequark: those D+/D- traces don't look nice
<wpwrak>
whitequark: you should keep them as short as humanly possible
<whitequark>
wpwrak: then I won't be able to do it on one side, which equals to not being able to do it at all
<wpwrak>
whitequark: also, are you sure the metal of your USB connector never touches the traces ?
<whitequark>
yes
<whitequark>
I've checked
<whitequark>
it has a bottom made of plastic
<wpwrak>
whitequark: (one side) worst case, you would need a wire
<whitequark>
oh
<whitequark>
I've leaved it plugged in for several minutes
<whitequark>
it's hot
<wpwrak>
whitequark: but you can probably do it as follows: rotate the chip so that D+/D- are very short. [...]
<whitequark>
well, the _working_ one I bought in store is exactly as hot as mine
<wpwrak>
(hot) that tells you something ;-)
<whitequark>
I guess that's a feature somehow
<whitequark>
all cp2102-s I've seen were hot as hell
<wpwrak>
whitequark: (hot as hell) including boards made by others ?
<whitequark>
wpwrak: yes, including perfectly working boards I bought
<whitequark>
and on some Digilent board, too
<wpwrak>
i can't remember the c8051f32x getting unduly hot, and the cp2xxx just seem to be variants of the same design
<wpwrak>
hmm. strange.
<whitequark>
still, it's hot.
<whitequark>
and I agree that's very strange
<wpwrak>
whitequark: anyway, back to D+/D-: rotate the chip. connect signal ground to shield ground. then you have ground on both sides, connected via the USB connector.
<wpwrak>
whitequark: for extra paranoia points, connect the two ground sides also on the other end, under the header
<whitequark>
hm
<whitequark>
I can't rotate the chip. the lines in cp2102 and the usb connector are reversed
<whitequark>
that wouldn't work in one layer
<whitequark>
anyway, thanks for suggestion
<wpwrak>
ah, right, they are. okay, option 1: use a reversed connector.
<wpwrak>
option 2: use a 0603 0R resistor
<whitequark>
hmm, the resistor
<whitequark>
yes, it may be an option
<whitequark>
I'll do the new pcb tomorrow I think
<wpwrak>
option 3: use a full-sized USB plug (or make a PCB plug)
<kristianpaul>
whitequark: hacking an atben? :)
<kristianpaul>
s/atben/atusb
<whitequark>
wpwrak: through-hole mini-usb plug will work, too, as it'll go to other side of pcb
<wpwrak>
whitequark: yes, if you can find one that actually does work ;-) have you used one of these yet ? how think are your PCBs ?
<whitequark>
wpwrak: by PCB plug do you mean something like very cheap, small and crappy usb flash drives?
<whitequark>
wpwrak: yes, I know where to buy some. my PCB is 1mm thick (0.8mm sheets often look bend in the shop, so I prefer to not buy them)
<whitequark>
kristianpaul: no. just trying to do a cp2102 usb-uart for third, or maybe fifth, month
<whitequark>
kristianpaul: that's going to become a project of my life
<whitequark>
wpwrak: (pcb plug) that'd help a bit some time ago when I was killing one usb connector after another with heat gun
<wpwrak>
whitequark: (1 mm) okay, that should be thin enough
<kristianpaul>
whitequark: oh, an fdti alternative, good :-)
<whitequark>
wpwrak: (projects) I've actially did 3 or 4 projects in that timeline :) and sold two of them (other two were for me)
<whitequark>
wpwrak: it's just the fucking cp2102.
<kristianpaul>
never needed an idbg :p
<whitequark>
kristianpaul: yeah. it looks good in the DS, it is cheap, it is installed on a lot of pre-manufactured boards
<whitequark>
kristianpaul: but when you're trying to do one...
<wpwrak>
whitequark: (heat gun) i see the words, yet i lack the comprehension
<whitequark>
wpwrak: ahem. I think I've seen a phrase constructed like that somewhere
<whitequark>
wpwrak: there is a tool for soldering smd components which generates hot air
<whitequark>
wpwrak: I've just called it 'heat gun'
<wpwrak>
whitequark: usb is very finicky. oh, and did you make sure all your fancy flux is washed off the board ? as in 3+ cleaning cycles with hot water and alcohol, using a brush ?
<whitequark>
wpwrak: of course it wasn't. I may wash it 3 or 5 times, but if I heat it a bit... the flux instantly goes from all the hidden holes
<wpwrak>
whitequark: (heat gun) okay, "hot air pen" or such then. "heat gun" is the construction-grade kind of device.
<whitequark>
wpwrak: and how is it called correctly?
<whitequark>
(flux) I've heard of a chemical called FLUX-OFF from Cramolin. it's said to remove traces of flux easily and completely. maybe I'll buy one tomorrow
<whitequark>
wpwrak: mine is the station one. particluarily, the big thing at the end of pipe is exactly the same
<wpwrak>
flux remover can help, yes. also picking a "friendly" flux should help. e.g., water-soluble. i wouldn't quite trust "no-clean". sometimes it does need cleaning anyway, and then it's harder to get rid of than water-soluble
<whitequark>
wpwrak: hmm, I do have water-soluble. it's called TAGS, and it is based on glycerine
<wpwrak>
okay. good station. yet unnecessary for such a simple circuit. just a temperature-controlled soldering iron would do nicely ;-)
<whitequark>
wpwrak: the station is different, through. mine has an temp-controlled iron too :)
<wpwrak>
(flux) as long as it's for electronics ... (no idea what they put inside)
<wpwrak>
that's why you clean with alcohol afterwards. plus, water eventually evaporates. what's more of a problem are salts. they're okay while dry but when the circuit gets moist for some reason (e.g., condensation), they'll start conducting.
<wpwrak>
(station) looks nice
<whitequark>
wpwrak: (flux) well, everything other I've tried works with Pb/Sn+RMA perfectly
<whitequark>
mostly atmel controllers with a few peripherals, but there were various circuits
<wpwrak>
RMA is a bit hard to remove. but yes, it has good wetting properties. better than water-soluble or no-clean
<wpwrak>
were the atmels qfn, too ?
<whitequark>
wpwrak: no, only tqfps. I didn't have a reason to pick qfn atmels
<wpwrak>
aha ! qfn is trickier :)
<wpwrak>
did you try to solder the center pad ?
<whitequark>
sure, it should be soldered per DS
<wpwrak>
well, if that chip gets so hot, you may actually need it ...
<whitequark>
they require a ground connection there
<wpwrak>
the center pad is tricky. trying to solder it can easily make things worse.
<whitequark>
hmmm
<wpwrak>
(ground connection) that's what they all say ;-)
<whitequark>
wpwrak: so, everybody lies?
<wpwrak>
(trying) the emphasis is on "trying". can cause shorts with adjacent pads, and you can get solder paste residues there as well.
<wpwrak>
oh, i'm sure the chips work better if you solder the central pad and do it properly ... ;-)
<whitequark>
wpwrak: I have a bright LED lamp, and the pcb is thin enough
<whitequark>
wpwrak: I can either look through it or look at reflected light
<whitequark>
in both cases I can see the soldered and unsoldered bounds of the central pad (the pcb polygon is a bit narrower than the pad itself)
<whitequark>
there's no shorts
<wpwrak>
dunno such things are easy to spot
<wpwrak>
okay
<wpwrak>
anyway, here's how i do qfns: set the soldering iron to a temperature where the solder liquifies conveniently. this is usually relatively hot. but better to go in hot and out again quickly than having to roast the contacts for a long time
<whitequark>
wpwrak: that's 300 for me
<wpwrak>
next, pre-tin all the pads. (apply flux everywhere, deposit a droplet of solder, then sweep with the soldering iron over the board)
<wpwrak>
then, wash off all the mess. (water+alcohol+brush)
<wpwrak>
next, flux the qfn's pads, place the chip, hold it down with tweezers, and apply solder all around. chances are that only a few pins will be soldered and there may be many solder bridges. that's okay. in the first round, the objective is to just make enough connections to keep the chip in place.
<wpwrak>
then, add flux again and add solder where needed and/or redistribute the solder. it's also good to do one side at a time, i.e., apply solder, do one pass, apply solder the next side, do that side, etc.
<wpwrak>
may need 2-3 passes until everything looks right. remove excess solder with a braid.
<whitequark>
hmm, I'll try this
<whitequark>
what flux do you recommend?
<wpwrak>
by then, all the "good" contacts should be detectable by visual inspection. particularly if using leaded solder and not having big ground planes.
<wpwrak>
then you can go around and look for shorts. it also helps to have "needle" type probes.
<wpwrak>
(flux) i use water-soluble most of the time. i find it a good compromise between resin (RA/RMA) and no-clean.
<whitequark>
wpwrak: what one? I may even order it from [name of the place you live in]
<whitequark>
just because I'll then get A Flux That Definitely Works
<whitequark>
ah yes, that's not an order number, but a product number
<wpwrak>
those fluke probes are spring-loaded. and they come with a set of tips: needle and "crown" (three tips). the latter is good to grab the edges of 0402 components and such. these probes are really great.
<whitequark>
wpwrak: 0402 is still too hard for me. I don't even know where I can buy 10's of those
<wpwrak>
ah, i use flux pens, not bottles ;-)
<whitequark>
hmm, flux pens... any ideas where can I order a pair of these?
<wpwrak>
digi-key ?
<whitequark>
in russia?
<whitequark>
hm, they actually have a working site digikey.ru
<kyak>
rejon: do you want to make it sound "slow" intentionally? :)
<viric>
Instead of thinking about a wireless nanonote....
<viric>
How far can a wired nanonote be from a internet-connected computer? :)
<kyak>
i guess there are 10 m USB extenstion cables
<kyak>
the question is if they would work
<viric>
Ah I meant *working*, yes. :)
<wpwrak>
rejon: i actually wonder where "Slowpan" comes from and if 6LoWPAN could maybe come from "Slowpan".
<zedstar>
slowpan...cool...although it does sound a bit like a grill that george foreman might market
<wpwrak>
zedstar: ;-))
<mat619>
hey there. i need some help with a HP nc6120 notebook. it doesn't power on anymore at all, regardless of AC power, battery in or not and being docked or not. any ideas where to start looking for errors?
<mat619>
hello?
<kristianpaul>
hi
<mat619>
hey there. i need some help with a HP nc6120 notebook. it doesn't power on anymore at all, regardless of AC power, battery in or not and being docked or not. any ideas where to start looking for errors?
<kristianpaul>
wrong channel, this is copylyeft hardware, try on #hardware please
<mat619>
ooops. okay! sorry and have a nice day!
<GNUtoo>
again...lol
<rjeffries>
wpwrak looking ahead, what will need to change to acieve greater range with ATBen and ATUsb?
<rjeffries>
in a future iteration, would allowing an external antenna allow it to work at greater distances?
<rjeffries>
wpwrak in NO WAY am I complaining your work is Most Excellent!
<wpwrak>
rjeffries: a better antenna could help, yes. although i didn't see much of a difference between the pcb antenna and the usual ~10 cm wifi antennas
<wpwrak>
rjeffries: (an external antenna also needs a connector and probably a cable, so you have a bit of losses there as well.)
<wpwrak>
rjeffries: the chip has more features, though: 1) it supports antenna diversity: two antennas and it picks the one that has the best reception. 2) it can drive an external amplifier. each such improvement needs one extra chip and some other RF components.
<wpwrak>
rejon: whee, nice ! maybe also mention that the first goal is just plain internet connectivity ?
<wpwrak>
rejon: the idea was to have a single design, with one side doing USB-to-SPI while the other did RF. then, one could either use it with USB, or cut it in the middle, and implant the RF-only part in a ben.
<rejon>
that is great
<rejon>
this is so cool
<rejon>
the sensor approach is great
<wpwrak>
rejon: then Ornotermes had the idea to use the 8:10 card slot for I/O (unrelated to wpan). I then made the blinkenlights board and UBB is an offspring of that
<wpwrak>
rejon: (sensor) which sensor ?
<rejon>
hahaha
<rejon>
i mean, that is what i'm most interested in
<rejon>
data gathering
<wpwrak>
rejon: (cont'ed) another offspring of blinkenlights is the design change on wpan: instead of one unified design, make two: one for USB, the other for the Ben.
<wpwrak>
rejon: one of the first ben-only designs were the one labeled "A" and "B" (100903). note that they didn't have a crystal. back then, i thought i could just let the ben provide the clock for the RF subsystem.
<wpwrak>
rejon: the other fork of wpan was the atusb board, first with an USB-capable MCU from SiLabs. labeled "2", "3" (20101229)
<wpwrak>
rejon: each of the designs had one major design change: i had to add a crystal to atben, because the ben's own clock was too noisy. you can see a first try in the middle of the image, labeled "20110115"
<wpwrak>
rejon: the one in the middle was a slightly reduced version of 20110115. the one in the lower right corner is the final result (or very close to it)
<rejon>
wpwrak can you drop some knowledge on that wiki page?
<rejon>
i like calling this SLoWPAN
<rejon>
that is a good to market
<rejon>
just like Slowfi
<rejon>
but slowfi is more about culture
<rejon>
term i came up with some time ago
<rejon>
we have all the domains on it
<rejon>
hahaha
<wpwrak>
(slowfi) oh, indeed :-)
<kristianpaul>
s/one/two
<rjeffries>
wpwrak what have we heard form Tuxbrain about the process of manufacturing initial run of AT-Ben and AT-Usb?
<rjeffries>
the suggested qty 1 prices of thes etwo wpan boards (Ben and US) add up to more than a Ben if my math is correct
<wpwrak>
rjeffries: not unless the ben has gotten a lot cheaper recently ;-) but i don't know his final prices. not sure if he does ;-)
<wpwrak>
rjeffries: (manufacturing) the pcb fab has sent back the panelized gerbers a few days ago. we sent them back some comments. not sure what has happened since.
<rejon>
if you can put some latest outcomes, pictures and more and we can get rjeffries to add some thoughts, great!
<rejon>
i tried to make simple
<rejon>
would love to add more innovation pages like this
<wpwrak>
pheew ... "a next generation wireless personal area network" seems a bit bold ;-)
<wpwrak>
rejon: it's more like "poor man's wifi for now, then see what else it's good for" :)
<kristianpaul>
:D
<rjeffries>
wpwrak assume all the atBen and atUsb units you shipped have arribed at their destinations?
<rjeffries>
wpwrak who besides wpwrak has been able to make a wireless connection using your devices?
<wpwrak>
rejon: "patent-free playground" may also be overstated. there are almost certainly a lot of patents around, too. it's just that the basic technology (chips, etc.) is more accessible than, say, with wifi. in the sense that you can easily buy chips, there are public data sheets, etc.
<wpwrak>
rjeffries: not sure if richard or tuxbrain tried. afaik, roh doesn't have a usable ben.
<rjeffries>
wpwrak 6LoWpan protocol is (I think) completley free and open, correct?
<kristianpaul>
smile for a wpan-sdr ;-)
<rejon>
wpwrak rjeffries please directly edit the page! :)
<wpwrak>
rjeffries: to the extent that anything recent can be completely free and open, given the current patent system
<rjeffries>
wpwrak did you send a set of atben atUSB to xiangfu (sp??)
<kristianpaul>
rejon: i dont bet werner will touch wikis ;-)
<wpwrak>
rjeffries: nope, only to adam (forwarded to tuxbrain), roh, and richard
<wpwrak>
kristianpaul: only occasionally and reluctantly :)
<rjeffries>
oh ok. Adam is looking at manufacturing aspects I assume?
<wpwrak>
rjeffries: that was the original idea. but now it's tuxbrain who will do the manufacturing.
<kristianpaul>
rejon: no, he is busy with mm1 RC3 i think
<rjeffries>
wpwrak I talked with an engineer friend yesterday and he's interetsed in seeing the atUSB and related docs, thinks he might repeat might have an application
<rjeffries>
wre there a few good links? I would then attempt to add themin Wiki to the stuff rejon recently crafted\
<wpwrak>
rjeffries: (advice) yes, adam is making suggestions
<rjeffries>
oh welll my OpenID credentials are not working for the wiki
<rjeffries>
wpwrak I CCd you on my email to John Baggs. he is one of the small group of gifted engineers typified by wpwrak.;)
<wpwrak>
rjeffries: thanks !
<wpwrak>
wikinnoyance: URLs that end in .ogv but are HTML ...
<whitequark>
wpwrak: hi again
<whitequark>
I have a weird question for you as electronics/wireless professional ;)
<whitequark>
I have a bluetooth headset, and sometimes I have a weird feeling that the samplerate for the sound is off for maybe 100Hz, e.g. it plays at 44000 instead of 44100
<whitequark>
somehow this only appears after it is interrupted by activity of other devices on the same bluetooth adapter, i.e. a mouse
<whitequark>
if that's true, how that can ever be (I assume that the sound samplerate is derived from the main crystal inside), and if it isn't, why it constantly appears and only with this particular headset?
<wpwrak>
whitequark: does the sample rate seem to stay off after the interruption ? or does it only briefly seem to change ?
<whitequark>
wpwrak: before the interruption it is OK
<whitequark>
and after it it stays for at least a minute
<whitequark>
and, given the fact there is no noticeable change, it either slowly returns to original value, or that's just I'm getting used to it
<whitequark>
the effect is absolutely reproducible, at least on one song I can interrupt it in a certain place and instantly hear the weirdness
<whitequark>
well, that still may be the human acoustic tract characteristics
<wpwrak>
whitequark: hmm, could be some error correction and/or buffering bug.probably not much you can do about it.
<whitequark>
wpwrak: there is around a second of buffer in this device. ECC... hm, how that can affect it this way?
<whitequark>
the interruption is as follows: first, it plays for ~second, then it interrupts for 0.8 sec, then it plays again for 0.3 sec maybe, and then it is silent again
<wpwrak>
i don't know the error correction BT uses, but it could be just some badly implemented interpolation
<whitequark>
the problem is that I'm not sure what does happen there exactly, and I'm curious about it
<wpwrak>
oh, the audio stream gets fully interrupted ? that's bad
<whitequark>
wpwrak: bluetooth audio is fully digital, and ECC is done on lower layers. also it uses some compression for audio
<whitequark>
I don't believe it can just be interpolated
<wpwrak>
well, some form of interpolation. not necessarily a straight line ;-) but you have all sorts of layered codings and such, where (small) data loss means loss of quality, not all-or-nothing operation
<wpwrak>
but if you really want to know where they screwed up, you'd have to find out what protocols and codings exactly are used there.
<whitequark>
I know that it uses A2DP
<whitequark>
hmm, as the first step I'll try to play sine, then record it and analyze
<whitequark>
*sine wave
<wpwrak>
yup, that's a very good idea. get the human factor out of the system ;-)