AK4395 vs. AK4396 listening comparisons

[sendler]

Output R

Gang, I suggest proceeding with caution in paralleling voltage output DACs such as the AKMs. There could be 'bucking' among the output pins - depending on the summing node implementation - potentially increasing non-linearity rather than reducing it.

It might help to use a seperate 100R resistor for each output leg to isolate them but now you are increasing the output impedence slightly which I have found to be audible.

 

[Radian]

Gang, I suggest proceeding with caution in paralleling voltage output DACs such as the AKMs. There could be 'bucking' among the output pins - depending on the summing node implementation - potentially increasing non-linearity rather than reducing it.

Well, one of the best Dacs in the world (APL) has 8 of the 4396 in parallel .

Greets,
Klaus

 

[YohY]

More to // dacs...

Well, one of the best Dacs in the world (APL) has 8 of the 4396 in parallel .

Greets,
Klaus

Yeah, but that's what I was afraid of, that there's more to this kinda mod,
but Alex isn't likely to share his expert knowledge on how to do this - an I wouldn't
blame him as he's selling a product based on this circuit design !

 

[Ken Newton]

Well, one of the best Dacs in the world (APL) has 8 of the 4396 in parallel .

Greets,
Klaus

Such performance may be dependent on the summing-node implementation utilized by APL. It's hard to say without a schematic. Also, the quality of sound which you say APL achieves may actually have little to due with paralleled DACs and much more to do with unrelated solution details. One cannot simply assume cause and effect.

Generally speaking, their are two objective improvements offered via the paralleling of voltage-output DACs. 1. Halving of output impedance with each doubling of DAC devices. 2. Reduction of noise by 3dB with each doubling of DAC devices (which also depends on having excellent power-supply isolation between each DAC chip). Balanced against those potential improvements is a POTENTIAL net degradation in distortion. Such potential distortion would be a function of how well the DAC outputs match. The situation would be different for current-output DAC chips.

The main point is that, IMHO, there is no automatic system improvement simply by paralleling DAC chips. Improvement is possible, but it may be minor. Especially, considering the effort involved with a proper implementation.