IRC log of #maemo-devel for Monday, 2010-09-13

« Sunday, 2010-09-12 Index Tuesday, 2010-09-14 »
*** shadsec2 has joined #maemo-devel00:20
*** shadsec2 is now known as ShadSEC00:20
*** _0x47 has quit IRC00:52
*** swc|666 has quit IRC01:24
*** SpeedEvil1 has joined #maemo-devel02:12
*** SpeedEvil has quit IRC02:14
*** SpeedEvil has joined #maemo-devel02:15
*** SpeedEvil1 has quit IRC02:16
*** SpeedEvil has quit IRC02:36
*** SpeedEvil has joined #maemo-devel02:52
*** SpeedEvil has quit IRC03:05
*** SpeedEvil has joined #maemo-devel03:06
*** swc|666 has joined #maemo-devel04:59
*** tealbird has quit IRC05:22
*** Milhouse has joined #maemo-devel05:47
*** kamui__ has joined #maemo-devel07:18
*** shinkamui has quit IRC07:21
*** achipa has quit IRC07:33
*** DocScrutinizer has quit IRC07:34
*** DocScrutinizer has joined #maemo-devel07:34
*** tealbird has joined #maemo-devel08:06
*** DeeGeneRaL has joined #maemo-devel08:15
DeeGeneRaLcan i get help reflashing my nokia n800?08:15
*** tealbird has quit IRC08:33
airwhat do you need?08:34
*** eggonlea has joined #maemo-devel08:49
*** boogeyman has quit IRC08:51
*** achipa has joined #maemo-devel09:20
*** swc|666 has quit IRC10:21
*** kulve has quit IRC10:58
*** amigadave has joined #maemo-devel11:02
*** kulve has joined #maemo-devel12:07
*** kulve has quit IRC12:26
*** kulve has joined #maemo-devel12:29
*** kulve has quit IRC12:31
*** kulve has joined #maemo-devel12:33
*** _kW has joined #maemo-devel13:04
*** _kW is now known as kaawee13:06
*** kaawee has quit IRC13:08
*** kaawee has joined #maemo-devel13:08
*** kaawee is now known as kawe13:09
*** MNZ has joined #maemo-devel13:13
*** lizardo has joined #maemo-devel14:00
*** kawe has quit IRC14:09
*** kawe has joined #maemo-devel14:20
*** kawe has quit IRC14:26
MNZI've had very little time for hackery this past few days, but I just sat down to work some more on the audio codec driver and I sort of have bad news :/15:00
MNZDocScrutinizer, SpeedEvil, ping15:00
SpeedEvil?15:00
SpeedEvilOh well.15:01
MNZThe de-emph filter is very limited, it has only one 'pole' and one 'zero'15:01
SpeedEvilYeah - I kinda assumed that15:01
MNZLong story short, amplitude response is only 0 at frequency 015:01
SpeedEvilWhen you have only 3 or 4 taps, the frequency response is not going to be complex.15:01
SpeedEvilIn practice, speaker protection doesn't require that15:02
MNZso with a high pass filter at cut off freq 250hz, with sampling rate 48kHz, amplitude is at about 50% at ~70Hz15:02
SpeedEvilAnd -3dB/octave should be just fine.15:02
MNZSpeedEvil, are you positively sure? Will this be sufficient?15:04
SpeedEvilNo.15:04
SpeedEvilUnfortunately, it's hard to tell without speaker datasheets.15:04
SpeedEvilI may at some point inthe nearish future be opening up my n90015:05
MNZI tried out the filter on headphone output with cut off at 200Hz and most/all of the bass was pretty much gone (as far as my ears could tell)15:05
SpeedEvilTo add a compass.15:05
SpeedEvilAnd could measure stuff a bit better at that point15:05
DocScrutinizeryup15:05
SpeedEvilI would be surprised if there was a limit other than total dissipation, and no clipping15:05
SpeedEvilThat is - under 70mW total (guess) and not actually banging against the stops15:06
DocScrutinizerspeaker amp is d-class without coupling C in the output, this means the amp is +-0dB down to DC15:06
SpeedEvilah.15:07
SpeedEvilThat is clearly one reason, and will be entirely fixed by a filter that doesn't pass DC15:07
DocScrutinizerit's probably rather easy to kill it with audio samples having DC offset or very low freq components15:08
MNZAnd the filter definitely doesn't pass DC15:08
DocScrutinizerwell, the amp input isn't DC coupled, but it has really decent capacitors in there, sufficient for freq down to some Hz15:09
MNZThis is how scilab plots the filter with cut-off at 250Hz http://i56.tinypic.com/21jvl2w.png15:11
DocScrutinizerC are 100nF, but amp inp Z is unclear (as I'm too lazy to search for datasheet :-P)15:11
DocScrutinizerwhat's X ??15:12
MNZmultiply by sampling rate (48kHz)15:12
DocScrutinizerI suppose it's not Hz15:12
DocScrutinizeraaah15:12
MNZ250Hz is at .005215:12
DocScrutinizerMNZ: I'd bet my speakers on this :-)15:14
MNZBy all means, go ahead :D Iike I said, I tested on headphones and all audible bass is gone.15:14
MNZIf you are willing to test I can roll a patch15:15
*** kawe has joined #maemo-devel15:16
*** lmoura_ has quit IRC15:17
*** lmoura has joined #maemo-devel15:17
DocScrutinizerhmm, what's meaning of "test" here. I don't think I'll go as far as trying to intentionally kill my speakers and see if that filter is preventing that15:21
MNZDocScrutinizer, so how far are you willing to go exactly :D15:23
DocScrutinizeryou *could* connect a DVM to headphones jack, and test filter response by sending a -6dB sweep form 0..1000Hz15:23
DocScrutinizerusual DVM should work on 1000Hz15:23
DocScrutinizersox e.g can create arbitrary sine waves at any freq and level15:24
MNZI'd thought of doing that actually, but for the PA side of things15:25
MNZie, play such an mp3 and dump PA output to see what it's doing15:25
MNZand by mp3 I meant audio file... they have become synonyms to me :/15:26
DocScrutinizerwell, do it for your filter as well, and I'll review and we both sign for correctness of procedure and results, so there's no more arguing about if that filter is good enough to replace PA15:26
DocScrutinizeryeah, but use .wav by all means15:26
MNZI wasn't *really* going to use an mp3 haha. Anyway, I'll see what I can do about this15:27
*** lupine_85 has joined #maemo-devel15:51
*** kawe has quit IRC15:54
*** kawe has joined #maemo-devel15:57
*** alvaro__ has joined #maemo-devel16:05
*** kawe has quit IRC16:22
*** kawe has joined #maemo-devel16:27
*** achipa has quit IRC17:08
*** edisson has joined #maemo-devel17:16
*** kawe has quit IRC17:35
*** kawe has joined #maemo-devel17:38
*** kawe has quit IRC17:46
*** amigadave has quit IRC17:57
*** kawe has joined #maemo-devel18:02
*** Milhouse has quit IRC18:02
*** Milhouse has joined #maemo-devel18:11
*** Milhouse has joined #maemo-devel18:11
MNZone lunch and some hackery later, the PA filter's cut-off is at 350Hz and amplitude only reaches 0 at DC18:27
*** jacekowski has quit IRC18:32
*** jacekowski has joined #maemo-devel18:32
*** pH5 has joined #maemo-devel19:10
*** shinkamui has joined #maemo-devel19:18
*** kamui__ has quit IRC19:22
*** _0x47 has joined #maemo-devel19:29
*** kawe has quit IRC19:33
MNZI've done my calculations and I've made the amplitude decline slightly faster than PA. I'm pretty convinced this is good enough for the speakers19:34
*** kawe has joined #maemo-devel20:00
*** wnd has quit IRC20:00
*** kawe has quit IRC20:06
*** mirr0r has quit IRC20:19
*** mirr0r has joined #maemo-devel20:19
*** _0x471 has joined #maemo-devel20:22
*** _0x47 has quit IRC20:25
*** _0x471 has quit IRC20:29
DeeGeneRaLcan i get some help with my nokia n80021:01
*** kawe has joined #maemo-devel21:03
*** DeeGeneRaL has left #maemo-devel21:11
SpeedEvilNo.21:15
*** kawe has quit IRC21:16
shadeslayerhehe21:24
MNZSpeedEvil, DocScrutinizer I've analyzed the PA filter like I already said and I set the deemph filter to a slightly more aggressive high pass. FrieT on #maemo is willing to test it for me21:29
SpeedEvil:)21:29
SpeedEvildon't you have a device?21:29
MNZI do, but I'm only 99% sure about this. That 1% = speaker blown. And my dad would kill me heh21:30
SpeedEvilah21:30
lcukMNZ, !!21:30
lcukheadphones21:31
MNZlcuk, yes I've tested on headphones, and it works21:31
MNZlike I said, 99% sure. It's really that I'm not sure if there's something more to it besides the low freqs that can destroy the speaker21:32
MohammadAG51hmm21:32
MohammadAG51someone I hate has an N90021:32
MohammadAG51will that do? :P21:32
MNZhehe21:33
MohammadAG51hmm, 1%? not good enough21:33
MNZactually, you can easily blow the speakers of anyone.... you just have to make them play something with mplayer directly to alsa21:33
lcukhrm21:33
MNZmplayer -ao alsa:device=hw:021:33
lcukthat sounds more and more worrying21:33
lcuk!21:33
MNZdon't do that kids21:33
MNZone catch though21:33
MNZthat's not enabled by default21:33
MNZneed to enable it in /etc/asound.conf21:34
MNZonly the pulse device is enabled21:34
MohammadAG51oh21:34
MohammadAG51wait, WHAT21:34
MNZso: add a few lines in /etc/asound.conf, then mplayer and boom21:34
MohammadAG51<MNZ> actually, you can easily blow the speakers of anyone.... you just have to make them play something with mplayer directly to alsa21:34
MohammadAG51wtf21:34
MohammadAG51doesn't nitdroid do that?21:34
MohammadAG51I've noticed a higher volume on it21:34
MNZthey probably have some other speaker protection somewhere21:35
MohammadAG51nope21:35
MNZor maybe that low freq thing is a myth....21:35
MNZ:D21:35
MohammadAG51at 100 percent the device kinda vibrates21:35
MohammadAG51cause of the loud sound21:35
MNZhrmmmm.....21:35
MNZdirectly to alsa? and there are no active alsa plugins in asound.conf that maybe used for filtering?21:35
MohammadAG51it vibrates like retarded bass21:36
MohammadAG51+a*21:36
MohammadAG51ask in #nitdroid, never tried it after a first run, too worrying21:36
MNZwtf. and the speakers worked again juts fine?21:36
SpeedEvilIt is unlikely to be instnat damage21:37
MNZwell if you are playing a 4 minute song that should be enough no?21:37
MNZ(I would imagine)21:37
MohammadAG51i played kryptonite twice and another song21:38
MohammadAG51more than 4 mins21:38
MohammadAG51speakers work fine21:38
MNZmaybe it wasn't bassy enough :P21:39
MohammadAG51or my head has awesome sound imagination21:39
MohammadAG51can't remember the other song, though it was bassy :P21:39
MohammadAG51i think it was Low21:39
MohammadAG51Flo Rida21:39
DocScrutinizerguaranteed speaker blowout21:41
DocScrutinizerscrew nitdroid21:41
DocScrutinizerrumour has it the last delay of N900 rollout was due to implementation of audio protection, as they found speakers 'burning' without it21:42
MNZBut if it's not instant damage then I should man up and just try it myself21:42
DocScrutinizerDon't think Nokia introduced this for mere shits and giggles21:42
* MohammadAG51 shits and giggles :P21:43
DocScrutinizerMNZ: it's unpredictable, and does wear your hardwrae no matter what, just like overclocking does21:43
MNZDocScrutinizer, I'm pretttty sure I've nailed it down though. The filter is now even slightly more aggressive than what PA is doing, just to be sure21:44
DocScrutinizersounds pretty ok to me21:44
MNZbtw, PA introduces a considerable bunch of nice to the audio stream21:44
DocScrutinizernice like what?21:45
DocScrutinizernice like renice 20 ?21:45
MNZnoise*21:45
MNZhaha21:45
DocScrutinizerooh21:45
ShadowJKatleast it's not as bad on N8x0 where the hw introduced massive amounts of noise :/21:49
ShadowJKbtw, I wish there was headphone blowout prevention in N900 :P21:50
ShadowJKI've blown a few when plugging in21:50
MNZ?????21:50
ShadowJKthere's a loud POP when I plug in headphones21:51
MNZon the N900 ?21:51
ShadowJKyeah21:51
MNZweird, never had that21:51
ShadowJKSometimes it doesn't do it if I plug it in very very fast21:51
ShadowJKI think the headset/tvout detection runs too early?21:52
ShadowJKor the amp is switched on too early.. or something like that21:52
MohammadAG51i've plugged mine in slowly/quickly21:52
MohammadAG51no pop21:52
ShadowJKor it gets turned on with volume at max21:52
MNZNo pops here either21:52
DocScrutinizerShadowJK: wtf? MIC_BIAS not disabled? AV-video-out enabled? headphones amp misconfigured?21:52
ShadowJKIt doesn't do it if I first plug in an extension cord ;p21:52
ShadowJKAnd the default nokia headphones don't do it either, but they have the sensitivity of a large brick anyway21:53
DocScrutinizerthis makes me think it's wrong about what is actually connected21:53
ShadowJKit does a few "biip biip bip" and presents the headset icon instead of headphone icon if I plug in headphones slowly ;p21:54
DocScrutinizernever knew there's a difference22:00
DocScrutinizeralso what's the bipbip?22:00
ShadowJKI thought I was hearing probing signals to detect headset button22:01
ShadowJKor something similar22:01
DocScrutinizeryeah, sounds like22:01
DocScrutinizerthough that shouldn't beep22:02
MohammadAG51i think it doesn't detect the button like that22:02
MohammadAG51that uses the 4th pin22:02
ShadowJKThese are the same class of headpones that can pick up the interference caused by N800 touchscreen signals22:05
ShadowJKand in the right room in my house, with the headphone coord dangling but unconnected, and the right shirt, I can determine whether there's power flowing to the water heater or not :D22:06
ShadowJK(there's a low pitched hum that builds up slowly and then pops and goes away when I touch the plug)22:07
MNZShadowJK, that is just too sensitive22:09
ShadowJKone notch above muted is a bit louder than my usual playback volume too :)22:10
*** silbo__ has joined #maemo-devel22:26
*** edisson has quit IRC22:35
airanyone know what mce command keycode locks maemo4?23:02
MNZDocScrutinizer, can I have a link to that alsa plugin that you told me about?23:21
DocScrutinizerMNZ: http://people.openmoko.org/joerg/ALSA/ACI/acihooklib/23:47
DocScrutinizerMNZ: see, this is the result of one year thinking and reducing complexity while exploiting what's already there23:52
DocScrutinizer:-D23:52
*** lizardo has quit IRC23:53
MNZNice doc :D23:53
MNZThis will come in handy23:53
DocScrutinizerthen, as nobody really picked it up, I abandoned it23:53
MNZit's in a usable state though?23:54
DocScrutinizerthought it'd be a brilliant concept, but nobody else was interested23:54
DocScrutinizerit works, needs cleanup23:54
SpeedEvilIntreresting.23:54
DocScrutinizertested with the .asoundrc you find there23:54
*** pH5 has quit IRC23:55
DocScrutinizeryou can call arbitrary commands on audio device open(), close(), and you can make open() fail when return code of called cmd is error23:55
MNZabout muting everything else, any ideas?23:56
DocScrutinizermeanwhile I hope the bug in ALSA we found and patched, was patched upstream as well23:56
DocScrutinizerMNZ: see softvol same .asoundrc23:57
DocScrutinizerthink about a softvol down after a dmix23:57
DocScrutinizeryou can ute *all* processes that had opened that dmix for audio output23:57
DocScrutinizermute even23:58
« Sunday, 2010-09-12 Index Tuesday, 2010-09-14 »

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!