it is not echoes, it is pre- postringing becouse frequency response of a very large steep filter (brickwall) will give ringing in time domain, analog filter with such frequency response will have the same ringing. Echo effects appear because frequency response not flat(passband ripple), for example DF with passband ripple +-0.00001db has echo amplitude -120db, for +-0.005db has -68db.
NOS DACs nowadays is popular simply becouse it is only a few good DF for parralel DACs and presence of DF provides an additional EMI and need very good decoupling and PCB tracing.
Look at this design http://diyaudio.org.ua/blog.php?u=2&b=13 ,proper PCB and very good DF (Passband ripple +-0.00001 db, Stopband attenuation 100 db,Multiplier 20*26 bit, Accumulator 45 bit, Lenght 213tap, NS,digital Offset) gives a sound to which NOS implementation is very very far
Last edited:
Hello Nazar_lv,
How could you reach this incredible accuracy? Since CD is recorded with 16bit, 96bd, 0.0015% error, +-0.001db pass band ripple, post processing could not obtain better. If we trust with DF algorithm, +-0.00001 db error is added to original signal. It is not the result of the global process.
CDDA has 16bits 44.1khz but this does not mean that we can get only 96 dB of S/N or not less than 0.0015% THD. Number of 16bits does not mean that we have 0.0015% THD, this only means that we have 6.02*16+1.76db S/N for uniform quantization, DAC/ADC with zero DNL and INL will have zero THD. Use of proper Noise shaping (such as SBM) wil increase s/n ratio on medium frequencies to -110...-130db
it is dependent on the ADC DF.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.