This review page is supported in part by the sponsors whose ad banners are displayed below

"I personally like our ARK MX+ very much. A lot of our domestic customers however think it too neutral and detailed. It sounds great on good recordings but exposes bad ones. People here are looking for a DAC that plays music as beautiful rather than as precisely as possible. What's more, personal tastes vary a lot based on feedback. For these reasons we decided to introduce the FFM flexible filter mode technology. This accommodates various tastes. I mostly use mode 1 but many of our customers like one of the other three. That confirms the technology's usefulness. What makes FFM different from other manufacturers' filters are our algorithms for different sample rates within the same playback mode. Our filters are individually optimized for each sample rate. I wrote a very detailed technical explanation on this here.

"One ESS9018 advantage is support for externally upsampled signal >1.5MHz. Even if one doesn't use external filters, it upsamples to very high frequencies based on the master clock. For some sample rates we still use our Sanctuary chip as digital filter and upsampler. For others we run the ES9018's internal upsampler but program its digital filter for uniform performance. By upsampling to a megahertz level the analog reconstruction filter can be very shallow. This exhibits higher slew rates and more linear phase to sound more analog.


"The ES9018 is a true 32-bit DAC with true 32-bit volume control. It won't compromise dynamic range like the earlier generations of digital volume controls did. Mark Mallinson already gave a detailed explanation of it in your prior Invicta review but I'll add a few things. First, to fully use the ESS volume control, the analog power supply of the chip must be an ultra low-noise design or the chip itself won't exhibit ultra-low noise. That degrades dynamic range. That's why we revised the DAC's analog power supply five times. Only the final version exhibited the extremely low noise which met our requirements. Such low-noise circuit design is very costly. Second, the post-DAC analog circuit too must be an ultra low-noise design or dynamic range in the preceding DAC is compromised yet again. To meet this requirement we changed the I/V and low-pass filter opamps from LM4562 to OPA1612. The latter's price duly tripled. In the final analysis I don't believe that current digital volume, even ours, compares to high-end analog types. But it's already superior to most ordinary preamps. If you're seeking true high-end sound or very steep attenuation rates, you're still better served with a well-designed analog preamp. Our Vega/Merak combo is deliberately based on the simpler-is-better paradigm for the highest possible value.
DSD IIR low-pass filters with 70kHz and 50kHz -3dB cutoffs with 3rd-order function
"The Vega's class A balanced output stage is the same as that of the Taurus Pre. It drives all manner of loads. Its output impedance is a low 4.7Ω with a max voltage swing of ±15V p-p and 100mA of peak current. This output stage easily drives 600Ω without any measurable distortion increase over a 10KΩ load. This is actually a dedicated preamplifier output stage, not merely a high-current DAC output.


"About DSD and the ES9018, we're dealing with a multi-stage delta-sigma chip. I think it converts 1-bit DSD to a multi-bit version but not 24/32 bits. Direct 1-bit processing of raw DSD doesn't seem possible as any DSP operation during post production, even the simplest summing or gain change, will request that the 1-bit data be converted to multi-bit. However converting such 1-bit data to multi-bit doesn't equate to DSD/PCM conversion. I did check the final output of Sabre's chip on DSD. That signal exhibited text-book ultrasonic noise typical of DSD which meant that it hadn't been converted to PCM. I also compared the same music on DSD and PCM. DSD sounded better. In the end we won't fret over exactly how the ES9018 processes DSD. The only thing of concern is how good DSD sounds over the Vega. For converters in general I think the presently most advanced is dCS's patented 5-bit RingDac. By the way, here and here are some lovely DSD128 and DXD samples from Kent Poon's Design w Sound blog. Simply register on his website to gain access to some free downloads which allow anyone to compare DSD and DXD directly."

For a brief return to magic numbers, comparing the group delay figures of AURALiC's four PCM playback modes shows how for 44.1kHz data, it progressively lowers from 794μs in mode 1 to 725 to 176 to 49μs in mode 4 whilst 352.8kHz data starts out with a far lower 99μs in mode 1 and diminishes to 40, 22 and finally 10μs in mode 4 which Xuanqian's paper calls the mode that during R&D received the highest beta-tester score. With the Vega it thus makes sense to experiment with 352.8kHz upsampling in software like PureMusic. For Mark Mallinson's claims about Sabre's embedded digital volume control performance, let's revisit his tables from my prior review.
16-bit CD analog digital
0 96 96
-5 96 96
-10 96 94
-15 96 89
-20 96 84
-25 96 79
-30 96 74
-35 96 69
-40 96 64
-45 91 59
-50 86 54
-55 81 49
-60 76 44
16-bit CD analog digital
0 96 96
-5 96 96
-10 96 96
-15 96 96
-20 96 96
-25 96 96
-30 96 96
-35 96 96
-40 96 92
-45 91 87
-50 86 82
-55 81 77
-60 76 72
24-bit data Analog Digital
0 132 132
-5 131 127
-10 126 122
-15 121 117
-20 116 112
-25 111 107
-30 106 102
-35 101 97
-40 96 92
-45 91 87
-50 86 82
-55 81 77
-60 76 72

The left table shows what happens when a CD is played through a DAC with a 104dB S/N ratio. As soon as we exceed 10dB of attenuation, the analog control begins to outperform digital. At -40dB the latter compresses recorded dynamic range by a whopping 32dB. The central table shows a DAC with a 132dB S/N ratio running Sabre's 9018 chip. Now digital volume control performance equals analog down to -40dB of attenuation and thereafter only trails by 4dB up to -60dB of signal cut. The right table displays Mark's "realistic assumptions" for 24-bit data when one begins with a 132dB S/N ratio