Open Source DAC R&D Project

[Kurt von Kubik]

well actually owen has managed -115+ DNR plus noise, using a modified pass mosfet D1 output stage with the sabre. this required rails of +/-50vdc though

I do not doubt that, but I think that both +-50V and FETs are out of scope.
We just need to do better than any of the following components with a comfortable margin, and we do need to enhance the ability to feed and control the next amplifier and the cables with both ease and elegance. Chasing extreme THD+N data far beyond what can be achieved in other necessary circuitry, makes no sense to me, besides the honour.
As already told, try out the very best op-amps, which actually can reach extreme data, and then compare them to a well build discrete analog stage.
Then you´ll know that we surely did not measure everything, regarding sound quality. There might be a few other issues to address to better SQ.

 

[qusp]

huh?? this WAS with a fully discrete analogue stage. not all that much better can be achieved with opamps, which is why I found it quite impressive and changed my build to accommodate it (considerable heatsinking and casework changes). the following stage in my case can actually achieve more than this. opamps in this following headamp were used in the super-regulator and servo only and my headphones are capable of 119db. thats the point, i'm not a huge numbers man, previously I was running transformers on the sabre, but doing this while not sounding like its trying too hard is not an easy task.

 

[esl 63]

For those of us with poor close-in eyesight there is some relief available in the newer 'reverse geometry' caps. Instead of being 0805 (about my limit for seeing the beasts clearly) they now also come in 0508 where the terminations are along the longer edges. These have a worthwhile decreased inductance than their corresponding 'widescreen' variants.

This is the way to go!
The self inductance is less then 1/5 on those reversed models for the same capacitance.
Also notice that the dominant inductance is the PCB path!
And since the geometry is reversed, the cap is narrower and can be placed even closer to the chip, which gives you even shorter path and much lower ESL!

Remeber where you read it first
I will solder 0306 caps directly to the DAC chip legs!

You can remember this too
And of course as big capacitance as you can ever find in that encapusling!
(have found 2,2uF, and this cap has approx 20 times less inductance than 0508 type which is common)

It cant be better than this I think...

 

[Ryssen]

Any progress in this project?

 

[esl 63]

Yes... unfortunately i guess som electrostatic discharge have damaged the player so i´m looking after a new one.
The reason is most likely that my cat was sleaping on the circuitboard... cat skin and IC´s is not a perfect combination, specially when the winter was extremely cold which lowere the relative humidity to a minimum.
I will do some more testing, disconnecting the boards again and then mount it.
I have put the TENT clock borad in 3.3 Volt clock mode. And I am powering the clock board up before i turn on the player itself.

Please Ryssen... if you have any ideas what to do you are welcome.
I don´t want to present the rebuild utill I get the player spinning
/Martin

 

[Ryssen]

No I don´t have any other Ideas than that tried in this thread:
http://www.diyaudio.com/forums/digital-line-level/137976-experience-diy-dac-380.html

I havent got the skills to do construktion work...

Edit:And I can add that I like the CS4398 mor than PCM1794.

 

[Ryssen]

I have read sommewhere on that clock problemI think it was with the DCX2496 but cant remember what the solution was.That having to power up before the player.
Maybee someone else knows.