Charles Hansen wrote:
I'm hardly a connoisseur of NOS DACs, but on paper it would seem that the PCM1704 would stomp all over the TDA1541. The low-level linearity, in particular, is far, far better on the Burr-Brown part.
No, not really. It is beyond question that the PCM1704 is excellent part and one of the bests in terms of linearity. However, when comparing the TDA1541A and PCM1704 papers / datasheets, you find the following low level performance @ -60dBFS: 0.79% for 1541A against about 0.85% for 1704. The PCM1704 is normally better when works with 24 bit data, 0.6% (which is, nominally speaking, a 17 bit linearity). Now, when you put this into the perspective and use dithered signal to overcome the limitation of 16 bit input of TDA1541(A), i.e. to overcome inherent limitation of non-dithered 16 bit signal itself, you get, depending on the actually used series of 1541(A) between 0.38% and 0.78% (this one for nominally worst, R1 "grade"). Actual graphs posted here:
viewtopic.php?p=13118#p13118 So, in my view, their performance in this regard is about the same, with TDA1541A having slight advantage, since the performance quoted for PCM1704, if I didn't misread the datasheet, is given for its best, K grade. However I didn't measure PCM1704 myself.