i've a doubt...
its often seen here that.. people going after realtime os or tweaking linux kernel to reduce system latencies...etc. is it to avoid jitter..? are they actually fighting jitter...?
reducing latencies is good for situations like in a recording studio where multiple tracks are recorded and played back simultaneously in sync. for example the feedback or tracks played into the earphone of an artist. if there is delay for a track being played to reach artists earphone.. the track being recorded will be out of sync with the played reference track (pilot). such strict timing is not at all required for just listening to a song. anyone can wait for a few hundred milliseconds before listening to a song.
according to my understanding (may not be correct nor complete):
jitter is the unevenness in the clock signal... coming to the DAC.
let me explain what i know about it:
every soundcard have its own independent clock source onboard... a crystal oscillator. the audio controller divides this clock to generate the 'sample clock' to the DAC. so the things which can cause jitter are the quality of the crystal oscillator and the electromagnetic interferences occurring to 'clock signal' at the PCB track between audio controller and DAC.
normally... no digital device can do mess with its own clock... it just runs on it. so no reason to blame the pci audio controller.
a busy system (cpu, os) can cause buffer underflow... missing samples.. discontinuity.. breaks in playback... such things. but it impossible to induce jitter... since there is no connection between the soundcard's 'DAC clock' and the system.
when u play an wav file the system does nothing at the rate of the audio's sampling frequency... it just reads data from the disk.. processes it and fills the samples to a buffer and informs (sets) its samplerate to the audio controller. its the audio controller's business to dig the pile of data available at its buffers and keep a continuous flow of samples to the DAC at the rate of 'samplerate'.. (eg. 44.1kHz)
imagine there is two conveyor belts... analog and digital. digital have slots (clock) and run by a motor which wows and flutters. analog is flat and runs exactly along real-world-real-time. what comes through the digital is transfered to the analog. the distribution of packets in the analog belt won't be evenly spaced. ie. distortion caused by jitter.
here the audio controller's job is to inserts packets into each slot of digital... it will not recognize any jitter nor can cause. the computer's job is to keep audio controller's basket full with packets (samples).. guess who causes jitter here.. and who is innocent.. :)
even an old pentium-100MHz can decode an mp3 and playback the resulting PCM data though a soundcard.. successfully..! so why it must be done in real-time at gigahertz beasts...?! :)
am i missing something...? is there something else which i don't understand in all those efforts u made to make it jitter-proof...? you are doing so much that it makes me doubt whether what i understand is right or wrong... or i do not know something which i really missed... or not able to understand... most probably all i have is just a simple understanding of the actual complex problem... :)
please let me know me why do so... what are u actually doing...
i started a new thread because the subject deserves it. (atleast for me)
It is interesting to read.. but difficult to digest:
You post looks interesting and I liked comparison with the two conveyor belts...
What you need to do now is start making audio kits, experimenting, trying all that the theory is telling us.... you'll realise that the sound card clock you are referring is fed by the Xtal, and all is powered from the same power supply. The amount of noise generated by other users in a standard computer system is huge. This noise will directly influence the jitter performance of the sound card in question.
In a perfect world we all strive to achieve to minimise the jitter, even the interference generated by the HDD seek movements of the reading head(s) mechanism will be enough for me to never consider a sound card as an audio source.
To me, the only digital system that can ever approach analog (turntable / records) is high quality, properly “encapsulated” in to the right plastics materials (Audiomeca Mephisto), CD drive mechanism feeding the simplest possible 1541 NOS DAC. Only here I heard the believable emotion in time. This system has only one jitter generator, and that is the read-out precision of the mechanism.
Thank you Boky...
will it help if i insert a passive LC filter to the power-supply lines in my PCI sound card...?
i search and read about Audiomeca Mephisto... it is interesting and it costs approx $7000..! LOL :) it is something which i see as 'gold plated technology'..!
all i have here at the moment is something below $1 (no bank account or any other dumps). and what i got here is some mp3's, a cheap sound card and a good amplifier built by myself.. still i'm a proud audiophile :)
see what i do now : http://www.ronybc.8k.com/ak4531-es1370.htm
It may help... I'm not sure how exactly you'll do that... cut the supply tracks on the sound-card PCB?
The trick with passive filtering is to cut the noise of OUTGOING glitches from other "noisy users"- not only on incoming power supply line(s)...
|All times are GMT. The time now is 07:18 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio