In
this review Miklorsmith said he didn’t have a preference between the Altmann Attraction and LessLoss 2004 DACs except that he would choose one or the other depending on which amplifier he was using. What I get from that is the two DACs are identical in regards to distortion, resolution, dynamics, and all the other audiophile attributes. I find that hard to believe. As I see it, the two DACs couldn’t be more different.
Just look at the datasheets. The TDA1543, used in the Altmann Attraction, is described as a “Dual 16-bit DAC (economy version).” Why “economy?” Philips compromised the design to reduce manufacturing costs. Read the specs. THD+N = -75 dB. That spec is typical for a 13-bit DAC. The ‘1543 throws away almost 20% of the low-level information on the CD! In
this thread I showed the poor settling performance of the ‘1543. Here is a
spectral plot of a 0dB, 1KHz sine wave. The extra peaks represent noise and distortion. (The spectral plot shows the frequency and relative amplitude of each component in the audio signal. A good DAC produces a single peak at the signal frequency,
like this.)
It’s unclear whether or not the Attraction DAC includes an anti-alias filter; most other TDA1543 NOS DACs do not. Aliasing is audible as high frequency, enharmonic overtones. Without the filter, what are essentially square waves are sent to your amp and speakers, which can lead to phase distortion and lots of other problems. Many audiophiles confuse phase distortion with increased sound stage. They are not the same.
On the other hand, the PCM1704, used in the LessLoss, is described as a “24-bit, 96KHz Sign-Magnitude DAC.” It is a much more sophisticated and costly design. The lowest grade chip has THD+N = -92dB, which is about par for a 16-bit DAC. That’s not so good for a 24-bit DAC but at least it’s not throwing CD bits away. Too bad LessLoss coupled the PCM1704 with a DF1706 digital filter. Digital filters are good at interpolating steady-state sine waves, which is what sampling theory is all about, but digital filters suck the life out of the music. Because I don’t have a DF/PCM1704 DAC handy, I’ll look at the output of two highly rated DACS with built-in digital filters, AKM and Wolfson.
Here is a
0dB sine wave eight octaves below the Nyquist frequency. The green trace is the AKM and the yellow trace is the Wolfson. Here is the
same signal as a two cycle burst.
This is what the signal should look like. Pre- and post-ringing is characteristic of digital filters. The post-ringing of the Wolfson is excessive. With both DACs, the frequency of the ringing is not related to the frequency of the signal and will be particularly obnoxious.
Here is a
one octave warble centered eight octaves below the Nyquist frequency. The AKM does a passable job but the Wolfson is all over the place. Notice the low frequency modulation and the forked peaks of some of the cycles. Digital filters just don’t do well with abrupt changes in the signal. Unfortunately, music is all about abrupt changes. Even the simplest melody starts, stops, and changes from one note to the next.
