Non-OS 44.1K residue

[dhaen]

Hi,

I'm just trying to decide whether to try a non-OS DAC.

What are the opinions on how low the residual 44.1KHz can be made without impacting on the audio passband (ripple, group delay etc)?
What is a feasible figure in dB below operating level, or better below dBF?

TIA

Cheers,

[ThorstenL]

Koinichiwa,

Originally posted by dhaen
What are the opinions on how low the residual 44.1KHz can be made without impacting on the audio passband (ripple, group delay etc)?
What is a feasible figure in dB below operating level, or better below dBF?

First, any Image up to around 40KHz is caused by frequencies normally piped to the tweeter and due to the sinc rolloff already notably down, as this rolloff reaches -3db @ 20KHz and is > -8db @ 30KHz. I have aspreadsheet to calculate the sinc curve for any funloving sample rate, but sadly even zipped it is > 3MB.

If the output waveform of the DAC is a propper ideal staircase than 40KHz as result of the sinc function will be down by 20db. However this "ideal" sinc curve will be blurred by non-ideal staircase (non-zero settling time of DAC) output.

Now, if you add my favoured "peaking" 3rd order lowpass you get an offset of the -3db @ 20KHz at the price of 90 degree Phaselag at 20KHz and a 3rd order rolloff above that, making 40KHz including the sinc function around -20db to -30db depending how fast the DAC settles.

IN MY PERSONAL VIEW such a level of attenuation of the first image suffices if well designed equipment is used. I'd even go as far as saying that the simpler LCR resonant Anti Sinc Circuit previously discussed with a 1st order LPF @ 44KHz combined with the sinc rolloff should be enough....

Sayonara

[AudioFreak]

Originally posted by Kuei Yang Wang
I have a spreadsheet to calculate the sinc curve for any funloving sample rate, but sadly even zipped it is > 3MB.

KYW,

If you email it to the webmaster account I might be able to put it on the server for you.

[ThorstenL]

Koinichiwa Webmeister San,

U GOT MAIL.

Sayonara

[AudioFreak]

I'm downloading the email now... It's time for bed so I'll put it up on the server tomorrow.

[ThorstenL]

Koinichiwa Webmeister San,

The mail failed.....

<p.....10@////.com>:
17.250.248.49 failed after I sent the message.
Remote host said: 552 5.2.3 Message exceeds maximum fixed size (3145727)

(the e-mail addy was altered to throw off spambots....)

I uploaded the zip to here for temprary parking:

http://groups.yahoo.com/group/Thunde...files/SINC.zip

Sayonara

[dhaen]

KYW,

If it's alright, I've put it here:
http://www.dhaen.org.uk/kyw/kyw.htm
I'll provide a more elegant link later.. e&oe

Jo mata

[ThorstenL]

Koinichiwa,

Originally posted by dhaen
KYW,

If it's alright, I've put it here:
http://www.dhaen.org.uk/kyw/kyw.htm

Yeah, that's cool. I simply wrote this on the quick a while back as I wanted to know "exact" values for a few thing in the context. As I could not find anything easy to use I made me one. Put it into the public domain...

Sayonara

[dhaen]

KYW,

I must say, the problem seems less than I'd thought.
The levels you quoted: were they below operating level? If so, which one Please tell me in dBF.

Okagesama-de,

John

(Nihon-go wa honto omoshiroi desu-ne)

[Pedja]

Hello Dhaen,

Originally posted by dhaen
I'm just trying to decide whether to try a non-OS DAC.

Try it, of course.

Originally posted by dhaen
What are the opinions on how low the residual 44.1KHz can be made without impacting on the audio passband (ripple, group delay etc)?

In fact it is impossible.

Originally posted by dhaen
What is a feasible figure in dB below operating level, or better below dBF?

Depends on what is the goal. Digital filters are at least 60dB (some the whole 120dB) effective and they attenuate practically immediately after the audio band. You can not achieve this by the analog means unless you destroy the sound completely. Fortunately, with decent rest of the gear, you actually do not need a filtering.

Check these pages, show the effectiveness of a few non-o/s filtering approaches (four are multiple notches, one is 3rd order low pass).

Pedja