How close will it get to the mighty KDAC? We will have to wait and see!
T, it sounds like the KDAC has been shelved on your system? Interesting. I guess there is no perfect component for all situations.
I might have missed it, but what is the technical reason why so many chips in parallel makes such a difference? Is it simply increased power on the output?
Paralleling dac's, resistors or just about any device is done to cancel random non linearities and noise.
The (simplified) theory is as follows: If you have, for example a voltage output from a valve and there is 1V OP and 1uV noise, the noise is random,
so if you parallel 2 valves, you get twice the voltage (or current) but since the noise on each valve is random then it will not add exactly,
some will cancel some will add. The general rule is the noise adds as the square rt x no of devices so for 2 devices = x 1.41.
So we end up for 2 valves: voltage = x 2, noise = x 1.41 so there is a net loss of noise.
As you can see the law of diminishing returns sets in, having to double the no of devices each time for a net gain of 1.41 (or 0.707, whichever way u look at it).
For DAC's there is a noise advantage but also some forms of distortion are also cancelled with multiple devices.
For the 1543, it is a MIXED scenario. It has almost 10 x the THD=N of the 1541. It doesn't specify how much of that is noise
so it's hard to predict what the actual measured improvement will actually be without running some tests on it, preferably
multitone IM. Even this won't shed much light on what the sonic result will be.
On thing is absolutely for sure, something like an ess sabre dac, when correctly implemented will measure better than even 1000 x 1543's in parallel
in every way that we currently know how to measure so - as usual, we are still some way off understanding completely how all this stuff
correlates to actual sound.
My theory is a big part of the 1543's 'sound' -is- it's distortion spectrum.
T