The job of the D to A converter is to decode digitally encoded signals back into analog signals that can be used by an amplifier. All DACs have four basic stages; a digital reciever that accepts digital data in various formats and passes it on to the actual conversion engine, the DAC chips or chips that do the actual conversion, digital filters which remove the artifacts of the conversion process, and finally the low power amplifier output stage. The biggest performance difference between different DACs largely center around the middle two stages.

Since the ground-breaking DAC 64, in conjunction with industry expert Robert Watts, Chord Electronics have configured their own conversion methods and implemented these at gate level within programmable "configure your own" FPGA chips. Although conventional "off the shelf" DAC chips have evolved so they are more suitable for the wide dynamic range of audio signals (on the order of a trillion to one) than they used to be, they are still constrained by leakage currents (and therefore power) that disrupt low level digital and subsequent analog signals. In fact as the building block architecture of chips - gates and nodes - become even more minaturized to now below 30nm, leakage current becomes a greater percentage of the total power managed by the chip (projected to reach as much as 50% of total power).

The D to A conversion process must be linear over the whole operating range for accurate re-assembly of the analog signal. The smaller the operating window that the DAC can reliably convert within, the greater the compression and distortion of the analog output signal which results in a reduced dynamic range (such as found with MP3 for example), less low level ambient information (lack of space, sound stage, depth and focus), and less timbral fidelity due to altered harmonic proportions (thin sounding brass, electronic piano sounds etc).

Linearity can be extended closer and closer to the limiting random thermal noise floor by making leakage power remain a very low percentage of the signal power. This was achieved 11 years ago by Chord Electronics with the ground breaking 64-bit DAC64 (most off the shelf chips are still only 32-bit) by using three massive Xilinx FPGA chips which offered many orders of magnitude more chip "real estate" for keeping signals better isolated. In 2009, Stereophile magazine awarded the Chord Electronics DAC64 mkII with the very rare and hard to obtain A+ status and partnered with the BLU upsampling CD transport received a product of the year award. Reviewers in the HiFi magazines raved this combination bridged the gap between digital and analog sound quality convincing many that the superiority of analog was now overstated.

Amazingly, shortly after receiving the award, Chord Electronics launched their next generation DAC, the QBD76. With the processing power of no less than four 70-bit DSP cores (interestingly cell phone manufacturers have just caught up with the 4-cored exynos chip) bringing "red book" CD performance close to the theoretical limit.
The QBD76 has a USB port for direct connection to a PC, and unique high quality stereo (A2DP profile) bluetooth wireless connectivity (Which preserves battery life of the transmitting cell phone or laptop far more than WiFi). A2DP, especially with apt-x, is completely different "hands free" and gives very high quality stereo sound reliably over a useful distance.

The QBD76's incredible sound quality is not just due to the D to A conversion engine but also the Watts Transient Alignment (WTA) algorithm 'FIR' filter used to remove the unavoidable side effects (like ghost images seen with broadcast signals on old TV sets) of the conversion process. Most analog or digital filters are not linear phase because they introduce a different amount of delay for different frequencies components of a signal. A FIR filter is phase linear; it keeps the delay the same for all frequencies, harmonic distortion is minimal, and since there is no feedback, impulse response is excellent (the trailing edge of a step response has hardly any ripple). Hundreds of millions of computations per second are required to implement this response hence the use of the Xilinx Spartan 3 chip. A true super processor!
The QBD76 also uses unique 8th order noise shapping (highest of any DAC) and 2608 times oversampling for better resolution and a smoother more natural sound. Jitter is reduced by the newly designed digital phase-lock loop (PPL) with 27 bit accuracy and a RAM buffer. Data-related jitter is now completely eliminated leaving only random master clock jitter of 3pS. The QBD76 uses a "three-dminsional" six layer circuit board and other advanced manufacturing technologies obtainable only at a few places in the world.

kemela © 2003-2012 | Privacy | Terms of Use