Audio, PA, XPROT, and the damage done
 

Indeed it seems it's XPROT and the gear around it that got invented (or at least adopted) by Nokia when it became obvious the class-D without highpass would blow their speakers, which now has some quite unpleasant effects on audio, like dynamic compression, dynamic EQ and other crippling of high quality audio.
Though all this might be needed and maybe even OK for the built in speakers, it's probably worth it to review and rework the sw-side audio chain when using headset connector, particularly when using it for 0dB line-out to any home-stereo or even professional equipment. Alas it's not possible to auto-detect whether headphones or a 3.5mm->2*cinch Line-In cable is connected, so any such adjustment has to happen by triggering it from a sort of GUI.

For first experiments it might be worth it to completely shut down PulseAudio and use the plain ALSA for audio output. CAVEAT: this scenario is known to potentially blow your speakers in N900.

PS: I don't think the class-D amp by itself is that bad.
Cheers
jOERG

Re: Audio, PA, XPROT, and the damage done
 

Well for the built in speakers then it is good to have it all there to allow for maximum output.
-I dont think that the way internal speakers are used is so demanding on quality.

I dont believe that headphones need any protection from the n900.
Headphones usually are capable to handle outputs from soundcards and are most likely designed to withstand other strong sources without breaking.

This would be very interesting to try on the headphones output ;)

But i was also thinking about if not speakerprotection could be made of passive electronics (a large cap in series with the speaker) or as d&b amplifiers which only uses gainreduction as speakerprotection but which has a sophisticated way to measure the temeraturedependant impedance of the speaker and to predict the movement of the cone. ( is conemovement, Xmech, even a problem here?) otherwise a amp with a very high outputimpedance (currentamp) could be used and a simple mathematical formula for the induced and transferred heat in the coil could be utilized in the digital world.

Re: Audio, PA, XPROT, and the damage done
 

They do that to protect your ears, not to protect the headphones ;-)

Re: Audio, PA, XPROT, and the damage done
 

Well, they dont know what efficiency (how much of the electrical power that is converted to sound and not to heat) i have on my headphones ;)
-I am familiar with the fact that there may exist some kind of recommended or maybe even legal limit somewhere in the world, but i think it is ridiculous since efficiency easily can vary 20-30dB between different headphones)

However, currentlimiting on the line-out would solve that.(a simple resistor?)
line input usually is in the 10Kohm range for modern pro equipment and more like 100K for homeequipment (Really old tubegear can be as low as 600ohm but i dont think anyone ever will connect a cellphone to that :p ) while headphones are more like 8-32ohm and a few professional models with as high impedance as 600ohm.

Ps, about you PM, i am quite convinced that it is the dsp i cant get my ears around, so nothing strange there.

Re: Audio, PA, XPROT, and the damage done
 

Yeah, I was going to say pretty much the same thing, except TMO was "under maintenance" and didn't let me :)

The phone can easily tell between line in and headphones by the impedance. I assume the same trick is used currently to tell between headphones-with-mic and TV out, but of course I don't know for sure.

Class D amp may be a PITA but class A or AB is a power guzzler and battery killer. Class A pre-amp for line out followed by class D for speakers and headphones?

Re: Audio, PA, XPROT, and the damage done
 

No, and no. N900 has no means to check speaker impedance. Test for headset, headphone, AV is done via "3rd ring" (mic/video) impedance. On this level headphone and line-out cable look identical.

Re: Audio, PA, XPROT, and the damage done
 

But Neo900 might. Or have I missed the mark? Is this not about Neo900?

Re: Audio, PA, XPROT, and the damage done
 

this is a leisure talk about audio. sure I will take ideas to Neo900 from here ;-) Audio is the most non-finalized section in Neo900.

Re: Audio, PA, XPROT, and the damage done
 

Regarding class A, that is a powerhog ;) But i dont think the crossoverdistortion in a class B bufferamp will be any worse than the stepdistortion from any cellphonefriendly D/A converter.

Anyway, limiting the output could be done by limiting the swing for 0dBFS too, thereby making it possible to optimize the electronics for any kind of capacitive/inductive output that our cables (and headphones) can present for the amp.
-Atleast i will always have somekind of preamp with variable gain to plug the output into ;)

However i still dont think it is useful to do anykind of "earprotection" when headphones dont have a standardized sensitivity.

Re: Audio, PA, XPROT, and the damage done
 

Check 1bit-DA converter, and the influence the clock rate has on precision. Actually some (most) class-D have no quantization at all. They don't work like e.g. 10bit parallel R-2R D/A-converters but like switched mode PSU, with either continually modulated pulsewidth at a fixed frequency, or fixed pulswidth at continually modulated frequency. Either way the modulation is done via analog means, like delta-sigma comparators. See http://www.beis.de/Elektronik/DeltaS...eltaSigma.html figure 1. (a class-D amp is a delta-sigma converter with analog output and usually even analog input)
Class B has most problems with the power dissipation in output driver transistors at I/2, when both current and voltage across a transistor C-E are at 50% of maximum. IOW when for a 4Ohm speaker the transistor has also "4Ohm" between collector and emitter, and both the speaker and the amp dissipate same amount of power. Class-D efficiency is theoretically 100% since either voltage or current is zero, dependng on switch (transistor) open or closed. Thus class-D in embedded.

full ACK for nonsense of ear protection.