Output stage design for DAC ?

[bordins]

Hi,

I am trying to find information how to design/select an output stage circuit for a DAC. Cheap players use opamps for the option while expensive players employ no-opamp (discrete) output stage. Some who love tube sounds put a tube in.

Some people said hi-end cd players do have the same circuit as what in hi-end preamps, in terms of sonic quality. I am curious what this circuit looks like.

How to design the output stage circuit to sound "better" than others ? Is there any theory behind ?

Thanks.

[audiodesign]

look my DAC 1.0 on www.audiodesignguide.com

[wa2ise]

My web page of analog audio tubes in CD players

I have modified several older players with tube outputs. From the DAC chips I usually do a passive IV conversion and use tubes to amplify it to line levels.

[bear]

From what I have gleaned, there are several ways to skin this cat.

Problem is, they all seem to sound different.

The opamp solution seems ok for the I/V part, and maybe even the filter, GIC preferred if an active filter is used. BUT the key here is that the opamp selection is critical. They are decidedly not all equal.

Some folks have decided to used "stacked" DAC chips, to increase the available current. This seems to have some merit. Trimming the DACs might be essential here - dunno, never tried it.

The passive "conversion trick" (just a resistor with an amplifying stage after) seems inviting, and may actually sound good, but since the current coming from the DAC chip is small, the signal can start to get into the noise. Here the stacked DAC chip idea has additional merit (but added cost and work).

Tubes seem problematic to me in the I/V position since they have significant noise of thier own. As a buffer, ok, I guess. Pick ur tubes wisely, though.

As far as buffers, I prefer a discrete buffer, but you may be able to use opamps well enough... especially some of the newer ones. I'd not give any consideration to the old BUF 03, imho.

The best thing to do, imho is to use two DAC chips (notwithstanding any stacking) run 180 degrees out of phase with each other, and use that to cancel completely all the common mode garbage that may or may not appear at their outputs...

Keep in mind that with the newest generation of chips, you need to reach down to a -120dB floor to get the full performance out of them - otherwise you're effectively truncating their performance by "injecting" noise...

Imho, if you have certain chipsets in front of a given DAC chip, it won't matter all that much what you do at the output end. Dunno why they seem to have their own "signatures" imprinted on the sound, of if these things are due to clocking issues or jitter or some other measureable parameter or not. Imho, a good example of this is the chipsets used by most of the Japanese mfrs... of course, ymmv.

Oh, if you've got a commercial player or DAC with opamps, the best thing (imho) to do initially is to try a range of BB/TI, LT & AD chips in those spots (socket the chips) and see what you hear... maybe change out the caps for better ones, and/or improve the PS filtering/regulation to the analog stage(s)... building from scratch is another thing entirely.

_-_-bear