#qi-hardware IRC log for Friday, 2011-08-26

qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-08242011-2111/00:41
qi-bot[commit] Xiangfu Liu: nanonote-files copy file to etc (master) http://qi-hw.com/p/openwrt-packages/6b6757300:45
qi-bot[commit] Xiangfu Liu: nanonote-files bump version number (master) http://qi-hw.com/p/openwrt-packages/0e59c0f00:48
wpwraktoday, i got a shipment from digi-key, containing three little lithium button cells. the warning labels on the box made it look as if it contained a nuclear warhead ;-)03:08
kristianpaulwpwrak: what you need lithium button cells ?03:28
kristianpaulself powered atben? :-)03:28
kristianpauldamn psu03:29
wpwraknaw, for powering the atben, i'd just feed back the regulator output into the input :)03:29
wpwrakit's to experiment with chargeable lithion cells. i basically want to prove they're not good at delivering the kind of power i want.03:30
kristianpaulif somebody here get are involved on making a laptop like thing with rechargable batteries, PLEASE avoid tampered PSUs...03:31
wpwrak(relatively high current, pulsed, but with fairly wide tolerance on the voltage. so the primary li cells are okay, but i have my doubts about the secondary ones)03:31
kristianpaulchargeable, sound nice :)03:31
wpwrakwhat's a "tampered" psu ?03:31
kristianpaulhe :p03:31
kristianpauli have a mips laptop, the psu is kinda small and easilly breaks when you try open it..03:32
kristianpaulso is no tampered.. sorry03:32
kristianpauland dont loose easy.. argh03:33
kristianpauli just need to change the broken power cable..03:33
kristianpauloh wel..03:33
kristianpaulif the thing will break why they just dont make it swichable, like the connector for the main..03:34
Action: kristianpaul looks for his coping saw03:38
kristianpaulfinally surender, the winner a scalpel :)03:59
kristianpaulmay be i'm not used right term, but this psu box was not actually glued04:00
kristianpaulsealed it looks like..04:00
kristianpaulbut the scalpel make trought :)04:01
wpwraksaves a few picocents. make a hundred million units and we're talking about real money ;-)04:01
kristianpaulsorry i lost the track..04:08
kristianpaulwas that about the cells?04:09
wpwraknaw, your cheap power supply04:09
kristianpaulah :)04:09
wolfspraulhey this is slightly off-topic here but then not really10:42
wolfspraulwpwrak: did you see the reports of people on the openmoko list about US carriers texting and calling FreeRunner users that their phone cannot continue to be used on the AT&T network?10:42
wolfspraulquite interesting10:42
wolfspraulI think we will see this more in the future.10:42
wolfsprauleven intercepting phone calls and redirecting the renegate users to call centers10:43
wolfspraulthe reason why I think we will see more of this is because networks are caught up in a jungle of old/legacy devices10:43
wpwrakis there an indication of the reason why they tell them to stop using it ?10:43
wolfsprauland have a hard time updating and supporting a myriad of devices with their basestations10:43
wolfspraulI think it's a general tightening of standads10:43
wolfspraulGSM is getting old, so the more old devices you can kick out the easier you can upgrade your network10:44
wolfspraulthe basestations10:44
wpwrakand where was this posted ? didn't even know the lists were up again10:44
wolfspraulotherwise the need to support a seemingly infinite number of legacy devices holds you down and creates support costs10:44
wolfspraulI've read some reports in that direction, but that's the first evidence from real FreeRunner users10:45
wpwrakfound it10:45
wolfspraulJoshua is also an avid NanoNote user :-)10:45
wpwrakokay, so they basically want to migrate users away from GSM. not FreeRunners per se.10:46
wolfsprauloh sure, they probably don't even know about 'FreeRunner'10:47
wpwrakthe last carrier supporting GSM may be in for a surprise, with all the FreeRunner users migrating to them and all of a sudden becoming a large group ;-))10:49
qi-bot[commit] Werner Almesberger: labsw/: schematics, layout, and header.fpd update (in progress) (master) http://qi-hw.com/p/wernermisc/c0bfec411:20
kyakxiangfu: so, the release image is probably being compiled right now? :)12:27
xiangfukyak, one small problem about nanonote-files. :(12:27
kyakyeah, a strange problem - didn't see it before12:28
xiangfukyak, yes. I setup a 'date' setup in uci-default . setup the date to the day of build when first boot. 12:55
xiangfukyak, that case that problem. 12:55
xiangfuI have update nanonote-files , using 'cp' not 'mv' may fix the problem12:56
wpwrakDocScrutinizer: btw, if you have a "LINE" input/output and a "TAPE" input/output, would you consider them to be compatible ?22:11
wpwrakDocScrutinizer: and are there any "standard" voltage levels for "LINE" ?22:12
DocScrutinizeryes, yes, iirc now 1V5 RMS, formerly 0V77522:12
DocScrutinizeror was it Vss?22:13
wpwrakVss ? or Vpp ?22:13
wpwraklet's see what wikipedia says, http://en.wikipedia.org/wiki/Line_level22:14
wpwrakwhee, a mix of everything ;-)22:14
DocScrutinizer>>Die Impedanzen und Pegel sind bei analogem Audio nicht einheitlich, so dass es bei Geräten verschiedener Hersteller, gelegentlich sogar bei Geräten unterschiedlicher Serien desselben Herstellers, zu Lautstärkeunterschieden bis hin zu verzerrenden Fehlanpassungen kommen kann<<22:15
wpwrakit's reassuring to know we live in a perfect world ;-)22:16
DocScrutinizer1V550 RMS22:16
wpwrakit's funny. my Sony stereo, definitely a CE device, outputs about 1.3 Vpp22:23
wpwrakso i wonder if sony added some extra oomph or whether wikipedia is a bit on the low side for consumer electronics22:24
DocScrutinizererr, how's 1.3 Vpp on the high side and 1.550V RMS on the low?22:25
DocScrutinizeractually 1.3Vpp sounds like a proper ~0dBV level22:26
DocScrutinizerinputs should cope with at least +6dBV22:27
wpwraki mean in reference to http://en.wikipedia.org/wiki/Line_level22:27
wpwrakaccording to the table, CE does only 0.447 Vpp, while this CE device of mine does almost three times that22:28
DocScrutinizerhttp://en.wikipedia.org/wiki/Line_level#Nominal_levels >>ARD, Germany  +6 dBu  1.550 (approximate)  2.192 (approximate)<<22:28
wpwraki'm still puzzled about what ARD have to do with _line_ levels ;-)22:28
DocScrutinizeranyway, I seem to recall it's been 0.775 in former times, and eventually changed to 1.5, of whatever unit RMS or Vp(p) or whatever22:31
DocScrutinizerand I'd think an input shall handle 2Vp while an output shall at least provide 0.5Vp22:32
DocScrutinizeras there's no strictly defined levels, and usually you always have a volume control somewhere, there's rarely an issue as far as I can tell22:33
wpwrakyeah. i'm trying to figure out if M1 does things right or not.22:34
wpwrakif you look at this guy, what input voltage do you think it wants ? http://www.wolfsonmicro.com/products/codecs/WM9707/22:34
DocScrutinizersometimes you find outputs that don't have Z  of 100R but rather ~1kR, those are a bit nasty22:34
wpwrakin what way ?22:37
DocScrutinizerwm9707 DS says >> Input level: (Z >10kR) AVSS-100mV .. AVDD+100mV << which I gather you'd consider Vpp for err probably sth like +9dB22:39
wpwrakWM9707 output seems to want 10 kOhm (page 5, Output level)22:39
DocScrutinizer10kR load is normal22:40
DocScrutinizersee above, it even specifies 10kR Z for own inputs22:40
DocScrutinizerwiki said same, me too22:40
wpwrakhmm. i wonder if there's a typo in the absolute maximum ratings, and it's not AVDD +/- 0.3 V but AVSS-0.3 V to AVDD+0.3 V22:41
DocScrutinizerZout=100  Zin=10k22:41
wpwrak(Zin/Zout) okay22:41
wpwrakokay, then we're good with lots of room to spare22:42
DocScrutinizerI guess you should, yes22:42
wpwrakcool. thanks a lot !22:43
DocScrutinizerInputs are scaled for AVDD eg; 0dBv at 5.0V is equivalent to 3.6dBv at 3.3V.22:44
DocScrutinizerADC Circuit Specifications (AVDD = 3.3V) 48kHz sampling \n SNR A-weighted (Note 1) 85 dB22:47
DocScrutinizerif you refer this to http://en.wikipedia.org/wiki/Line_level#Nominal_levels the you got something like +12dBU headroom, and your S/N at consumer audio -10dBV was 85dB - 22dB22:51
DocScrutinizerif I'm not mistaken22:51
DocScrutinizeron "proper" equipment with some ~1.xx Vpp-ish line-out, you rather have a S/n ~80dB which should be good enough22:52
wpwrakyeah, i think M1 usually doesn't care so much about audio quality anyway. at least not in the current uses.22:54
wpwrakaudio is mainly used to modulate video effects. nobody will see a few dB ;-)22:54
DocScrutinizerif you want to do it the professional way, you get a RCA receptacle that's floating GND, and an OPAMP plus some birdseed and a potentiomenter, to lowpass-filter -3dB@24kHz and balun the RCA in, and highpass-filter@5Hz22:55
DocScrutinizerooh, so I could've saved to type the prev line22:56
DocScrutinizerwpwrak: hey, you know you should spend another 1kR series, plus 2 clamp diodes, for the line-in when they go to WM-codec?23:01
DocScrutinizerand you need sth like a 100nF...1uF decoupling C23:02
DocScrutinizerand a 50k to GND after that C23:02
DocScrutinizerand a 50pF parallel to that 50kR23:03
wpwraksure ?23:03
Action: DocScrutinizer should look into WM9707 DS for application notes ;-D23:03
wpwrakM1 has an 1:2 input divider (2 x 6.8 k), then a 1 uF DC block23:03
wpwrakwhere did you find them ? :)23:04
wpwrakfig. 18 "RECOMMENDED EXTERNAL COMPONENTS" keeps things rather simple23:04
DocScrutinizermake the (de)coupling C 470nF ;-)23:05
wpwrakhmm, i somehow don't think anyone will go for that at the moment, unless something really serious pops up23:06
DocScrutinizerand seems the chip has internal pulldown R so the 50kR also is moot23:06
wpwrakbut indeed, the data sheet says 470 nF23:07
DocScrutinizerso what's left is are the 2 clamping diodes, a 50pF parallel to one of them, and a 1k in front of that towards the RCA receptacle23:07
DocScrutinizers/ is//23:08
DocScrutinizermaybe even a bead between the clamp diodes + 50pF junction and the C16/1723:10
DocScrutinizerRF on a ADC input can give you really weird effects ;-D23:11
DocScrutinizerand ESD is a must23:11
DocScrutinizerso you want the diodes and the R, the C, and probably the bead 23:12
DocScrutinizerI'd actually place C17 directly after the input, then the 470R...1K series, then the 2 diodes and 50pF, then the bead to pin2323:13
DocScrutinizerRCA-in -> C17 -> 470R -> [A] -> bead -> pin24; [A]-> diode -> VDD; [A]-> 50pF -> GND; [A]-> diode -> VSS23:17
DocScrutinizerAVDD, AVSS23:18
DocScrutinizer>> The devices performance with AVDD = 3.3V is shown in Electrical Characteristics. In 3.3V analogue operation, mid-rail reference scales to 1.65V. All ADC and DAC references are 2/3rds of their nominal 5V value. Input and output signals that are 1Vrms in 5V applications, scale to 660mVrms in 3.3V applications. It is not possible to produce an undistorted 1Vrms output by using the mixer PGA gain whilst running from a 3.3V analogue 23:20
DocScrutinizersupply. <<23:20
DocScrutinizerI did an error above, when I assumed symmetrical AVDD-GND-AVSS of +/-3V3, and based my dB estimations on that23:23
DocScrutinizeryou can tell from >>  [A]-> diode -> VSS <<23:24
wpwraklet's see if there's ESD ...23:29
wpwraknyet :-(23:29
DocScrutinizernot on the chip. Chip never have ESD for 'real life'23:29
wpwrakyes yes, i know what you mean ...23:29
wpwrakit's jack -> 1:2 vivider -> DC block -> chip23:30
DocScrutinizerhow'd you do a 1:2 divider there?23:31
DocScrutinizersound like trouble23:31
wpwrak6k8 serie, 6k8 to GND23:31
DocScrutinizerthe chip's input is specified as Z>=10k23:32
DocScrutinizerthat's not a 1:2 then23:32
wpwrakAudio GND is now equal to digital ground, after a bead connecting them made the whole box go belly-up when i connected it to my stereo :)23:32
wpwrakah, okay, not quite 1:2 :)23:33
wpwrakadam will have kittens ;-)23:33
DocScrutinizermore like 1:323:33
wpwrakyeah, three-ish :)23:33
DocScrutinizeractually (1/((1/6.8)+(1/10))) : ((1/((1/6.8)+(1/10))) + 6.8)23:36
DocScrutinizeror I messed it up23:37
DocScrutinizer[2011-08-27 01:37:21] <DocScrutinizer> (1/((1/6.8)+(1/10))) / ((1/((1/6.8)+(1/10))) + 6.8)23:37
DocScrutinizer[2011-08-27 01:37:22] <apt> 0.37313432835823:37
DocScrutinizerbut you never know, chip Z can also be >>10k 23:38
DocScrutinizeris's just specified to be >=10k23:39
DocScrutinizertoldya a devicer there sounds like trouble, esp in the 1:2 ... 1:10 range23:39
DocScrutinizeranyway a 50pF parallel to the 6k8 load of the divider shall suffice for RF rejection and ESD23:41
wpwrakso no varistor, clamp, or zener ?23:42
DocScrutinizerit's an analog input in the end of the day23:42
wpwraklekernel: are you reading ?23:42
wpwraklekernel: sounds like a worthwhile addition for rc423:42
DocScrutinizerI guess they have some minimal clamping in chip anyway23:43
DocScrutinizerwit another 12kR in series this should survive OV23:44
DocScrutinizerwell 6.8 series, 6.8 parallel23:45
DocScrutinizermaybe increase the 50pF to make a lowpass 3dB@20kHz23:46
DocScrutinizerso it eats more energy of any surge23:46
DocScrutinizerI gather it's a mere signal analysis input for controlling purposes? then the divider issue also is rather irrelevant as you'd get away with sth like 20dB S/N there23:49
DocScrutinizeryou'd not want to listen to it, but for controling some parameters of a GFX algo it's just fine23:50
DocScrutinizeryou possibly even would *want* to cut frequencies >10kHz for this purpose. Results in less aliasing concerns23:51
Action: DocScrutinizer waves23:53
DocScrutinizermeh, wolfson had great whitepapers about e.g analog vs digital ground and how to design layout for that. Vanished from their site23:58
DocScrutinizerI don't think beads were mentioned at the one pont where both connect ;-D23:59
--- Sat Aug 27 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!