AURALiC VEGA DAC Review Highlights
The AURALiC VEGA is a fantastic product that has surprised me to no end. Looked at objectively on the test bench it offers performance that is unmatched. Listening to it subjectively it reveals more of the music than anything I have on hand. It is a true reference product from a company I hope continues this level of performance.
AURALiC VEGA DAC Highlights Summary
- PCM support to 32-bit, 384 kHz/sec and DSD128 support
- Toslink Optical, AES and USB inputs
- Balanced and unbalanced outputs
- Effortless, pure sound
- Intuitive design and interface
- Matching streamer coming soon
Introduction to the AURALiC VEGA DAC Review
A year ago if you asked me who AURALiC was, I could not have told you. Since then they have been all over the Hi-Fi press, with their AURALiC VEGA DAC being omnipresent. Though I saw it everywhere, I didn’t read any of the reviews as I was too involved in other things to worry about it. When the AURALiC VEGA arrived at my door one day, I had no idea what to expect from it. I even let it sit there unopened for two weeks as I was in no rush to get to it.
AURALIC VEGA DAC REVIEW SPECIFICATIONS
- Design: Solid State DAC
- Inputs: Optical, 2 x Coaxial, AES, USB
- Outputs: RCA, XLR
- Format Support: PCM to 32/384, DSD128 over PCM (DoP)
- Display: 512×64 Pixel OLED
- Size: 2.6” H x 11” W x 9” D
- Weight: 7.5 Pounds
- Price: $3,499 USD
- SECRETS Tags: AURALiC, VEGA, DAC
Since I unboxed it and connected it to my system it has remained in constant use. The AURALiC VEGA is a fantastic product that has surprised me to no end. Looked at objectively on the test bench it offers unmatched performance. Listening to it subjectively it reveals more of the music than anything I have on hand. It is a true reference product from a company I hope continues this level of performance.
A DAC is a hard component to buy and show off to your friends. It’s a box that sits there and operates in the background. It’s not a big TV you can hang up on the wall, or a subwoofer that you can rattle their teeth with. But while the VEGA might not be flashy, it produces a musical experience that will resonate long past the 20Hz sound waves of a gigantic subwoofer.
Design and Setup of the AURALiC VEGA DAC
The AURALiC VEGA DAC has a very clean design. The front panel has a single OLED display, in a comfortable yellow color, with a knob that controls it. Pushing the knob awakens it from sleep (standby), while holding it down returns it to sleep. A single press brings up a menu system you can navigate. Being a DAC, there is not a huge selection of items to edit, but you can select your inputs here and adjust the filter modes.
Turning the knob outside of the menu system adjusts the internal volume control. It moves from 0-100 in 0.5dB increments. Bench testing indicates that leaving the volume at 100 produces the best measurements and for all my testing I left it there. You can run the VEGA direct to an amplifier and use this volume control but a preamp provides better performance. The unbalanced outputs from the VEGA operate at over 4V maximum, not the standard 2V of CD. Most preamps will be fine as long as you turn the volume down a bit. The package includes a remote, but I never used it.
Inside the VEGA uses the ESS Sabre 9018 DAC. The VEGA upsamples all incoming data to 32-bits and 1.5 Megahertz before sending it to the Sabre DAC. According to AURALiC this enables a better analog stage and gets the best performance out of the DAC. There is no option to disable the upsampling and both PCM and DSD material go through the process.
Another key feature to the AURALiC VEGA DAC is the custom clock. AURALiC says their clock is accurate within 82 femtoseconds. The ESS Sabre is already good at reducing jitter without extra circuitry. This clock, combined with upsampling, reduces phase noise from clock errors to below -168 dBc/Hz. There are three clock modes to select from and I used the ‘Exact’ version for my testing.
The AURALiC VEGA DAC is in my system with a Marantz AV7005 processor, Emotiva XPA-1 monoblocks and Revel f208 speakers. I used unbalanced connections for all components and the volume control of the VEGA was at 100 for all listening. The main source component was a 2013 MacBook Pro laptop running J River Music Center 19. If you’re using a PC, you need to install a USB driver for high-resolution audio but not with a Mac.
The AURLiC VEGA DAC In Use
For many years now I’ve used the same selection of tracks on every single component I listen to. Even if some of the tracks are not audiophile grade, my experience with them lets me hear what a component does right and wrong. I’ve heard the tracks so many times that I have a good idea of what I believe they should sound like at this point. Then I listen to them on the AURALiC VEGA.
The most astonishing thing I heard from the VEGA are details in the tracks that I hadn’t heard before. It’s an audiophile cliche to say “I never heard that before”. Often it is that you haven’t listened to that musical passage recently. I listen to these tracks 2-3 times a week and still manage to hear details from the VEGA that have eluded me before.
As “Carnival” from Natalie Merchant draws to a close, the fine background noises are easy to distinguish. They don’t cover each other up, or vanish behind other noise, but are distinct and clear. Her voice is clear and clean, free of any graininess at all.
“Let Down” from Radiohead has drums in the background that often get covered up by everything else in the song. Through the VEGA I hear Phil Selway’s drum work even with the bombastic attack mixed on top of it. The clarity of the cymbals and other instruments is first rate here. Thom Yorke’s vocals resonate just a bit better than I’ve heard them before.
More stripped down material shines with the VEGA. Individual instruments sound more natural than through my usual reference, the Oppo BDP-105. The decay of a piano, or the strum of a guitar, have a smoothness to them that isn’t usually there. There are no artifacts or harsh bits to the music, it just sounds pure. This pure, natural smoothness makes it easy to listen for hours. Even with the volume levels pushing the limits of my room it doesn’t fatigue me over time. Instead it just makes me want to listen to more and more music.
Listening to the DSD download of Dave Brubeck’s “Time Out”, the VEGA is fantastic. It handles the signal over USB using the DSD over PCM (DoP) format. Again the clarity of the instruments on their own is wonderful. During the quiet passages I can hear the sound of something squeaking in the background, like a chair on a wooden floor. The squeak of a musician rocking back and forth during the recording is one of those details, inaudible before, that the VEGA reveals.
There isn’t a real downside to the AURALiC VEGA in use. It’s simple to setup and use, sounds glorious, and supports everything. OK, there is one downside: connecting over USB drains my laptop battery fast but AURALiC has a remedy for that too. They just announced a new companion for the VEGA, the Aires wireless streaming bridge. With the VEGA it will allow access to your whole music library from an iOS or Android app.
Notes from Editor (JEJ): I also had a chance to listen to the AURALiC VEGA DAC in February of this year. I was going to review it, but something came up, so I forwarded it to Chris for review.
I concur with Chris’ subjective listening in that the VEGA performs in an exemplary fashion, and, in my opinion, more in keeping with Ultra-DACs (costing more than $10,000). In particular, I could hear the sheen of cymbals when listening to my favorite jazz albums I obtained by high resolution downloads (most of my jazz albums are on vinyl, but a few I have in digital format). This sheen is a characteristic that could only be heard if the DAC has first rate low-level linearity, which it does (from the bench test results). If the linearity had changed direction (bending and then heading upward – see the linearity graph in the bench test section) at a higher level, say – 90 dB, the cymbals would have had a slightly harsh edge to them, instead of the silky smooth sound that they had with the VEGA.
Full orchestra also was very impressive, with no midrange mush (a charicteristic of low IM distortion), and low level detail was easily distinguishable, especially using headphones, which is due to the low harmonic distortion of the VEGA.
For the price, this is one of the best DACs on the market. – JEJ
The AURALiC VEGA DAC Review On The Bench
Analysis by David A. Rich.
All measurements were made using Audio Precision dual domain system test system that is on loan from the company. The measurements are made by Chris Heinonen. No schematic was made available. The VEGA is said to use patented technology, but I was told they were filed in China only. The website mentions circuitry from Neve 8078 professional mixing consoles, but an examination of the discrete circuits from that console on the web showed nothing that would have produced anywhere near the noise or distortion performance this product delivered.
The website does not give a DAC model number but many reviews sited the ESS ES 9018 hidden under a metal can. The performance of the VEGA is limited by the performance of the DAC. The challenge is developing analog electronics have that preserve the DAC’s performance.
Preamplifier and Power Amplifier Selection Requirements for a DAC with State-of-the-Art Performance
The VEGA has Signal to Noise Ratio filtered from 20Hz to 20kHz of 122.5dB ref dBFS in the worst channel. That hardly changes A-Weighted to 124.5dB ref 0dBFS. The small change is a result of very small power supply noise and a flat noise floor at low frequencies, as we will see below.
The most significant issue with the product is it has a full-scale output of 4.2VRMS to achieve maximum SNR at both the RCA and XLR outputs. Going lower, the noise floor of the analog section would dominate, but this just kicks the can down the road.
The preamp has to shed the gain before the signal does to the power amp. Assuming a typical power amp gain of 20V/V (29 for amps set to THX specifications), the amp would have to supply 880 Watts average into 8 Ohms and 1760 Watts average into 4 Ohms to reproduce a 0dBFS (full scale signal out of the DAC) at 4.2VRMS. A THX approve amp would have to clip after 1850 Watts average into 8 Ohms.
The lowest noise power amp we have tested had an SNR of 101.1dB ref 1 Watt at 8 Ohms and a maximum output level of 320 Watts (0.1% THD) into 8 Ohms. That is an SNR at 126.15dB, ref 320 Watts. The input level for this THX approved product at 320 Watts is 1.75VRMS. So we have to reduce the gain from the VEGA by 0.42 (-7.6dB) with a volume control that will produce an SNR of 126dB (cascaded with the power amp the SNR would be 123dB). This will match the VEGA. I do not know of a preamp with this SNR.
Here I am only dealing with the issue of getting the SNR to the speaker. The room noise would make it impossible to hear the noise from the VEGA unless 0dBFS produced a level so high it would produce hearing damage.
Closed back headphones would get you a lower noise floor, but the issue of the full-scale level being dangerously high does not go away. This is not a problem with the VEGA. Any DAC that produces 124dB SNR is going to exceed what the human ear can hear on the low end (I cannot hear that -110dBFS tone, so could you turn it up) or the other (0dBFS is way to loud at 120dB SPL, so turn it down).
I am not worrying about the distortion of the DAC. The distortion of the speaker, especially if it uses passive crossovers, will swamp the distortion in the electronics.
All tests, unless otherwise noted are for a signal with sample rate of 192k samples / sec, 24bit depth and Filter 1 (more about filter options below).
The DAC performance is the highlight of the unit. The -90dBFS sine wave has almost no spurs (tiny noise peaks). Note it is twice the size normally seen since dBFS is doubled to 4.2VRMS
The line linearity plot shows nothing but the noise floor. The line is at 0dBFS, which is 4.2VRMS until the noise dominates. The curve is up 3dB at -123dBFS.
Zooming in shows insignificant non-monotonic behavior of the curve. This graph stops at -120dBFS, so we never see the 3dB point. The offset at 0dBFS is not real, but a measurement issue. The AP is normalizing on the left side of the graph instead of the right.
Looking at the -60dBFS high-resolution spectra shows nothing but microscopic hum spurs (-135dB). Those hum spurs show up only because this is a high resolution FFT. They will be below the noise floor in the standard FFT below. Note this graph above is in units of dB relative to full scale (dBrA). Thus -60dB is 4.2mVRMS. A spectra at -90dBS was equally text book level (not shown). I will again emphasize the noise floor of a spectra is not the SNR. We did this on high resolution looking for spurs above the noise per bin of the FFT. Note that no noise shaping is visible out to 40kHz. With some cheaper DACs we have seen the noise shaping.
Increasing the amplitude to 0dBFS, which is 4.2VRMS or 12dBV, which is the unit on the Y-axis for the graph below, shows the largest spur -120dB down. The spur is a millionth of the signal level ! !
Looking at this spectrum from 1kHz down shows not a single hum spur. Remember, the signal is at 12dBV so the spurs are 12dB lower than the number on the Y-axis. The only signs of a discrete spur are all at -140dB down, at frequencies whose origin I cannot identify. A measurement at -140dB is 1 in 10 million or 0.00001%
THD + N at 4.2VRMS (0dBFS) was -114dB (0.0002%) it went up at 2kHz, -110dB (0.0003%) and stayed that way until 10kHz. The THD was -106dB (0.0005%) at 20kHz. In the 19kHz / 20kHz DIM test had all intermodulation products below -125dB (0.0006%) except the spur at 21kHz which was -122dB.
To bring in a perfect 10 the crosstalk at 10kHz was -130dB.
Digital Volume Control Option
The digital volume control function is internal to the ESS ES 9018 and not a special function added by AURALiC.
The digital volume control works by multiplying the digital word by a factor less than 1. Looked at another way, as the gain is reduced, the MSB bits stay at 0 (-6dB volume setting is a multiplication of 0.5), and then the next bit is dropped out as the volume is further decreased. When reducing the digital volume to -20dB (multiplication of 0.1), the top 3 bits are not moving with the fourth less active than it would be at full volume. Since the noise of the DAC is finite, we have given up 20dB of SNR.
We tested this expectation. The VEGA has an A-weighted SNR of 125.1dB ref 4.18VRMS. When the VEGA digital volume control was set to -20dB, the SNR was measured at 105.04dB ref 418mVRMS A-weighted. The 0.06dB difference from theory is likely experimental error.
A DAC chip is operating at optimal performance when the volume reduction is done in the analog domain. The analog preamp drops the signal and the noise level of the DAC chip. The digital control drops the signal level only, but the noise from the DAC chip stays constant.
As I discussed above, the VEGA puts demands on the analog volume control that are slightly outside the state of the art. We did not have a state of the art preamp for testing, but we did try the AURALiC TAURUS headphone amp which has a unity gain noise of 115.5dBFS A-Weighted ref 2VRMS in its better channel for the RCA line out.
Luckily, the noise level dropped when the volume control was set for -20dB. It appears the TAURUS has some analog active circuitry before the potentiometer, and this circuitry noise was reduced with the volume setting.
We attached the VEGA to the TAURUS and set the gain of the VEGA full scale (4.18VRMS) and reduced the gain of the TAURUS by 20dB. The output from the TAURUS was 419mVRMS since mechanical pot had limitations on how accurately it could be set. The VEGA – TAURUS cascade returned an A-weighted noise of 105.2dB ref 419mVRMS, which was a tad better than the VEGA alone (measured above at 105.04dB), so the TAURUS did move the DAC noise before its own noise dominated. To show a larger noise improvement would have required an analog preamp with lower noise.
Even with these closely matched results, I prefer to get the best SNR out of the DAC chip and to stay as far away from the noise floor of the DAC chip as possible by using an analog preamp. The DAC chips noise floor may not be flat as a result of the non-linear oversampling process. The analog preamps noise is dominated by a flat thermal noise.
It is interesting to note that DAC boxes with ESS ES9018 DACs from Oppo and Benchmark both choose to use an analog volume control instead of the digital option in the ES9018.
The VEGA has four digital filter options. Each trades high frequency flatness and reconstruction tone attenuation for different impulse responses. Filters 1 and 2 are similar as in the frequency domain as are filters 3 and 4. Kids and dogs might get unhappy with the reconstruction tones coming up in level by perceptual masking should prevent that. An identifiable sonic change from the impulse response (not shown) and group delay (not shown) has not yet been shown in blind tests. AURALiC provides a very detailed presentation on the filter types and the performance in the different domains in the URL below:
As can be seen in the document, significant changes are identifiable only for a sampling rate of 44.1k samples/sec. Our measurement of the frequency response Filter 1 at a sampling rate of 192k samples/second is shown below:
The response of filter 3 at a sampling rate of 192k samples/second is shown below:
Filter 1 Frequency Response at 44.1k samples/second is below. Note that this plot is limited to 20kHz.
A spectrum of 20kHz digital input at 44.1 k samples/sec with filter 1 selected is shown below. The primary reconstruction tone is at 24.1kHz and is down 120dB. Note this graph is again relative to full scale so 0dB, which is 0dBFS, is 4.2VRMS. This graph is also on a log Y-axis.
For those of you who worry about clock noise and tones note, there is no sign of phase or frequency modulation around the 20kHz tone at 4.2VRMS.
There is no untoward surprise in the time domain, full scale at 0dBFS of 4.2VRMS or 5.9V peak. All one can see is the pure sine wave with all the other spurs below 120dB. You really cannot see distortion in a time domain waveform below 1%, but I have a reason for mentioning this because of what I am about to show you.
Filter 3 shows a significant roll off at 16kHz but below that things are much flatter than some competitive products with non-standard filters.
For filter 3 the change in the level of the first reconstruction tone is dramatic with an input of 20kHz and a sampling rate of 44.1k samples/second. The 20kHz tone is down 3dB, and the first reconstruction tone is down only 17dB down from the 20kHz tone.
Looking at the impulse and square wave responses of the different filters for something that might result is a subjective change is almost like looking at tea leaves. Instead, let’s look at the 20kHz and 24.1kHz reconstruction tones interact to change the way the time domain response looks likes for a single 20kHz digital sine wave input. A so-called beat tone has formed at 4.1kHz, and the effect on the amplitude response is shown in the plot below.
The time domain plot was only taken to 150µsec. To show the full beat tone, I did a computer simulation to show the repetition of the pattern using the amplitudes shown in the above spectra. In the X-axis the grids are 200µsec and the Y axis is 2V
Even for kids (who actually can hear 20kHz) this should not be audible as a result of the 20kHz larger tone masking the 24.1kHz tone, which is not spaced far enough apart to hear as two tones. Note no non-linear effect is being shown here, just the interaction of the two sine waves.
Filter 4 was similar, with the rejection of the 24.1kHz reconstruction tone increased to 20dB, and produced similar beat tones.
Clock Acquisition and Phase Noise
DAC box manufactures tend to make a big deal of clock recovery performance. The AURALiC VEGA gives a phase nose spec of phase noise of -168dBc/Hz, but that is really not useful.
The change in phase noise from the center of the clock frequency would provide some useful information:
Walt Kester “Converting Oscillator Phase Noise to Time Jitter”
AURALiC cannot resist adding a jitter spec, which is also equally meaningless without context. It gets further clouded because of the ASRC built into the ESS chip.
Robert Adams “Clock Jitter, D/A Converters, and Sample-Rate Conversion” pp 10 -21,33, 44.
The VEGA appears to use PLL clock recovery designed to produce low phase noise despite the Asynchronous Sample Rate Converter (ASRC). The ASRC should get a fixed clock which can have a lower phase noise that one that needs to “pulled” to match the clock frequency coming in from the SPDIF or USB source. A trade exists between the ability of a PLL to lock for the purpose of clock recovery and the ability to provide a clock with lowest phase modulation and discrete spurs. The acquisition process can be aided by using different type of PLL phase detector, a change in the PLL loop bandwidth or the PLL loop order for acquisition. After the system has locked the PLL is reconfigured to produce a clock with less phase noise.
If the clock recovery system cannot maintain lock, an adaptive system can change the parameters and configuration, but the recovered clock will not have the same properties such the level of phase noise at a given frequency offset from the carrier. AURALiC provides no information on the clock recovery system used but it is adaptive.
It has three levels of adaption depending on the incoming digital data. This is outlined in the owner’s manual. Again it is not clear why they are going to all this trouble with for the recovered clock given the ASRC that will follow.
I am making a big deal of this because the unit allows the user to override the adaptation process, but as the manual says, if you do not let it adapt, it may not lock. This is going to worry people needlessly. They system is designed to work well with any incoming clock from a functional device.
Bench Test Summary
The VEGA is as good as it gets, especially at this price point. Perhaps the use of an ESS ES 9018 in mono mode with clever analog summation circuits following might get 1dB better SNR. It is hard to imagine anybody producing a unit with lower harmonic distortion, intermodulation distortion, and hum spurs.
Conclusions about the AURALiC VEGA DAC Review
I avoided hooking up the VEGA DAC for a few days because reviewing a DAC is hard. I always worry I’ll hear things that aren’t there, or hear no difference. Writing 1,000 words about not hearing a difference would not be much fun, so I delayed the pain. Of course with the VEGA I found myself having no such issues as it has been a pure joy to listen to.
At $3,500 the VEGA is an expensive component, but it more than lives up to that price. On our objective bench tests, it performs better than anything I have tested before. Subjectively, it provides a clear sign of that engineering with a refined, natural sound. I will venture to say this is the best sound I have heard in my home theater room to date.
For some of the best digital sound out there, you must audition the AURALiC VEGA. It is a reference class product, and I can’t wait to pair it with the upcoming Aires bridge so I can listen to even more of my music with this kind of sound quality.