Review of the Audio-gd DAC19 DSP & C2 amp - The ACSS connection
For those wondering what ACSS means it is the acronym for Audio-gd Current Signal System which is similar to Krell’s CAST system.
What it means is that when you connect your whole chain in ACSS, there are 2 interesting things:
1. The signal is being transmitted in the current domain, which minimizes the effect of the cables on the transmission.
2. The whole system acts as one single gain unit, and there is only one I/V (current to voltage conversion) being done at the output stage.
That way the system can avoid multiple gain stages and impedance mismatches.
Before getting the dac19dsp and c2, I used to own the older models (the dac19mk3 and the c2c head amp). What pushed to sell the older ones and to acquire the newer models was curiosity rather than dissatisfaction with the previous units. I was curious to see how the DSP-1 digital filter performed against the PMD100 and DF1704 digital filters that were available on the dac19mk3 and also I was curious to try the ACSS connection against the classical RCA interconnects to see if there is any change.
Describing the units:
I decided to write this in-depth description of the units for 2 reasons:
1. To give an insight on the design philosophy and choice of components in the dac19 and C2 as there are a lot of interesting things going on inside those units.
2. To answer pre-emptively some technical questions that usually get asked about the innards of the units, and how to connect them through RCA and ACSS.
Those who don’t usually like the tech talk can probably skip this part and jump right to the review part where I left my subjective listening impressions.
The DAC chip – PCM1704uk:
While it is true that in entry level DACs the power supply and the general topology will affect most the sound, I believe it is quite different once you go up the ladder. In more resolving devices, where every thing starts to matter, it is very hard to make a good sounding (and resolving) DAC from any chip.
The DAC19 uses 2 multibit PCM1704uk which are more than 10 years old now. Why would a company use that chip? Well, according to many respected audio figures, it is the best 24 bit dac chip ever produced.
A few years ago, DAC makers discovered they could make their chips a lot cheaper by abandoning the expensive to manufacture multibit chips (TDA1541, TDA1543 ...) in favour of the cheaper to make sigma-delta chips. What they didn’t realize at that time is that those sigma-delta chips generate a lot of amount of out band noise and sound horrible. Fortunately, Burr Brown/TI kept making (in small numbers) the PCM1704 which is as far as I know the only remaining R2R/multibit dac chip in production today.
The following article (see here: Mother of Tone - Conversion Techniques) explains in more details the differences between multibit and sigma-delta chip. If you look at the high frequency sine waves produced by sigma-delta converters, you will probably understand why so many people were complaining about the sound of digital audio.
Most high-end manufacturers use the PCM1704 for their top models (Naim, Esoteric, Wadia...) while they use sigma-delta chips for their “entry level” gear...
So why doesn’t everybody use them? For one, they cost a lot: You need at least 2 PCM1704 to make a single ended stereo DAC (and 4 for a balanced one). While most “high-end” sigma delta chips retail for no more than 10-20$ for a stereo dac, a pair of PCM1704 costs usually more than $70 ... when you can find them. And the PCM1704 chips need an external digital filter to make things even more complicated and expensive.
The dac19mk3 came with 2 swappable digital filters: the PMD100 and DF1704. The PMD100 was a rare and discontinued HDCD capable digital filter that was limited to 24/55 (but experiment showed it could go as high as 24/88). The DF1704 was a 24/96 digital filter and was used with the slow roll-off feature in the dac19mk3. Both of those filters are considered among the best external filters and have much better characteristics than most built-in digital filters that come with sigma-delta chips.
However, because audio-gd could not source anymore the discontinued PMD100, they had to make a new unit in order to allow the user the choice between the DF1704 and the DSP-1 used in the higher end models from audio-gd.
In this case, the digital modules are not easily swappable and require soldering according to the designer: so one has either to go with one or the other.
So what is the DSP-1? It is a very powerful 250mhz Altera Cyclone chip which sole purpose is to “condition” the data before it reaches the PCM1704uk. It has a whole lot of functions including: user configurable oversampling settings (from NOS to 8x), Three-stage linear-phase (with zero group delay distortion), PLLs to reduce jitter...
Also, since the dsp-1 treats the audio signal as parallel data and not real time streams, it is supposed to be less sensitive to jitter than traditional filters such as the DF1704 and the PMD100.
Analog filtering/Output stage:
All the analog filtering after the PCM1704uk is done in the current domain and is done by discrete components.
The DAC 19 is an all discrete design, DC coupled design. It avoids opamps and capacitors in the signal path which is very rare nowadays.
Overall, the DAC19 has very short signal paths and doesn’t use any negative feedback whether in the filtering stage or at the output stage.
The DAC19 has 2 sets of outputs: the ACSS and the RCA. While the ACSS keeps the signal in the current domain, the RCA output adds an extra step buffer by converting the signal from current to voltage before outputting the DAC.
The power supply is a key element when designing a DAC. A good power supply can help reduce the jitter and help the digital section perform close to its theoretical limit.
The DAC19DSP uses a beefy 50w R-Core power supply. I have seen many so called “high-end” DACs and CD Players with ridiculously small power supplies inside big shiny boxes (no more than 10/20w). I am not even talking about those using cheaply made switch mode power supplies...
By looking at the internals, you realize that a big chunk of the space inside the DAC is occupied by the power supply (transformer, capacitors, regulators ...) which is always a good sign for a digital component.
Here is quote from the website: “The DAC-19 uses 8 groups voltage regulators which are combined into double parallel voltage regulators to lower the internal resistance and to purify power supply. A separate Silence control circuit is powered with each regulator separately. Such power supply setup is rare in domestic designs and it inherits substantial prestige and advantages of imported high end products.”
Overall, the designer seems to have taken seriously the issue of power supply. However, it doesn’t mean that the dac-19 is totally immune to the power supply. Its designer acknowledges that its performance could be further improved by using good quality power filters and power cords (which has also been confirmed with my subjective listening).
The DAC-19 uses the DIR9001 which has a much lower jitter than the CS8416 but is limited to 24/96. According to the specs sheets the DIR9001 has better jitter rejection with a low 50ps figure, while the CS8416 has a higher figure at 200ps.
By the way, I was able to compare both digital receivers on the same DAC and can confirm that the DIR9001 sounded better and cleaner.
Curiously enough, the only DAC on the audio-gd line where you can choose between the CS8416 and DIR9001 is the lower end FUN model. For the others, starting from the DAC19, it was judged that using anything else than the DIR9001 would be a sonic compromise. So the 24/192 capability is “sacrificed” in favour of a better effective sound at the more commonly used rates of 16/44 and 24/96.
The usb input is just an afterthought (in my opinion) and uses off the shelf PCM270x chips. It is limited to 16/48 and there is nothing earth shattering about it.
Until asynchronous usb chips become widely available those usb inputs can be considered (in my experience) as temporary means of connection. For the best sonic performance it is advised to use the optical or the coaxial input. A good spdif connection (whether optical or coaxial) can galvanically isolate the computer from the DAC. Most usb connections do not provide any galvanic isolation which can cause noise pick-up problems at the dac level.
The only companies that have decent DACs with good asynchronous usb inputs are Ayre, DCS and Wavelength. However, those DACs are usually pretty expensive.
The C2 headphone amp/preamp:
The C2 has 3 inputs: 2 RCA inputs and 1 ACSS input. The RCA and ACSS inputs have different signal paths.
The RCA input has to go through a buffer stage. The stock opamp is the OPA604 but audio-gd offers different discrete opamps (I ordered mine with both the Earth and Moon opamps).
The gain setting is effective only on the RCA inputs and you get to choose between the +4db and +13db settings on the front panel.
The ACSS is a more direct path. The signal doesn’t go through the buffer stage and is not affected by the gain settings on the front panel. When connecting the dac19 with the c2, both units act as a single one and there is one gain stage at the c2 level. This avoids the redundant gain stages and impedance mismatches that happen in the classical RCA connection.
Current gain and volume control:
As described for the dac19, the C2 uses the current gain technology, which allows audio-gd to have a very linear circuit design without resorting to negative feedback.
The volume control is very different from conventional systems (whether using traditional pots or stepped attenuator). Here the signal doesn’t go through the volume pot but rather the volume pot acts as a reference for the current gain. So the C2 is in fact an amplifier with variable gain.
The C2 amp has a similar oversized power supply philosophy as the dac19. The C2 uses a 50 watts R-core transformer (which is better suited for audio than toroidal according to many designers), and a lot of separate filtering.
While the C2 has some nice features (discrete design, current gain, zero feedback ...) it doesn’t reach its full performance until it is used from the ACSS connection and paired with the dac19.
Main Chain used for this review:
Foobar v1.0 --> Kernel Streaming --> Jkeny’s modified hiface --> Oyaide DB-510 --> Audio-gd DAC19 DSP --> Artisan Cables "Ultimate Silver Dream" RCA / Sharkwire ACSS --> Audio-GD C2 (Opa Earth and Moon) --> Artisan Silver Cables --> Sennheiser HD-650
Power related accessories:
Hi-Fi Cables & Cie PowertransPlus Power Cords (x2)
Hi-Fi Cables & Cie SimplePower
Bada LB-5600 Power Filter Plant
Essential Audio Tools Noise Eater Parallel Filter
E&T Spider Rack, Maple platforms, Sandboxes, Brass cones, Acrylic and Fiber carbon sheets, Yamamoto footers, Herbie's Audio Lab Tenderfoot, SuperSonic Component Stabilizer
Headphone cables: Artisan Silver Cables Mark1, Artisan Silver Cables Mark2, Revelation Audio Labs Cryo Silver Sennheiser Cable
Transports: Emu 0404 usb, Musiland Monitor 01 usd, Teralink X, M2tech Hiface (stock), Purepiper usb to spdif, Teralink X2
DACs: Emu 0404 usb, Zero DAC, Audio-gd DAC 100, Audio-gd FUN, Purepiper DAC A-1
Headphone amps: Audio-gd ST-3, Little Dot MKIII, Audio-gd FUN.
Pictures of my system:
Some of the reference tracks used for this review:
Natalie Dessay - Italian Opera Arias - Emi Classics
Mahler - Symphony n 5 - Decca
Puccini - La Boheme – Decca
Mozart Violin Concertos - Marianne Thorsen - 2L - 24/96
Sol Gabetta - Schostakowitsch Cellokonzert Nr. 2/Cello
Vivaldi - Concerto for 2 violins - Carmignola/Mullova
Keith Jarrett - Paris / London - Testament - 24/96
Glenn Gould - The Goldberg Variations 1981
Hans Zimmer - Gladiator Soundtrack
The Dave Brubeck Quartet - Take Five
Diana Krall - Live in Paris
The World's Greatest Audiophile Vocal Recordings - Chesky - 24/96
Norah Jones - Come Away With Me
Patricia Barber - Companion
Soundrama - "The Pulse" Test CD
Head-fi/Chesky Sampler - Open Your Ears - 24/96
A little background on my listening preferences:
About one year ago (before getting the dac19mk3), I was seriously starting considering going to vinyl. Since I listen to a lot of classical music, I started to get tired to listening to violins and pianos that sounded digitized. I went auditioning a few DAC and CD players but wasn’t satisfied. The more revealing the DACs, the more flawed they seemed to me: for a moment I thought that when choosing a DAC, you had a sort of slider where you could either increase warmth or resolution but never both at the same time.
However, since going analog would have complicated a lot my life (I know nothing about setting vinyl), I took my chance and ordered the dac19mk3 with the PMD100 and it was a true ear opener. I gained both in naturalness of sound and details compared to my other DACs. It completely led me to reconsider what I had believed until then to be the limitations about digital reproduction. The dac19mk3 didn’t have any excessive false warmth (like some tubed output CD players and DACs) but it retained the natural warmth of the recordings, which made it very addictive to listen to.
Listening a few months to the dac19mk3 “educated” my ears and made me very sensitive to distortions made by sigma delta based DACs. In a “blind” test, I listened to 2 Audiomat DACs in a system I wasn’t familiar with. At the time of the listening, I had no knowledge about what was inside those DACs. After only a few minutes I preferred the Tempo 2.5 over the Tempo 2.6. The owner of the 2 DACs wasn’t a technically inclined guy and didn’t know what was inside neither. So I got curious, and as soon as I got back home, I did a little bit of research and found out the 2.5 was using a multibit PCM1704uk chips while the 2.6 was using a sigma delta PCM1794 dac chip.
This little digression is just intended to give the reader a little background on where I stand about DACs. Because of the type of music I mostly listen to (classical), I am very sensitive to the naturalness and accuracy of timber of a DAC. And I have found so far that only R2R/multibit dac chips can provide that naturalness and accuracy of timber. On the other hand, I am more forgiving about headphone amps. If I had to choose between a good DAC and a poor amp or the other way around, I would always favour the first option.
Hopefully, this little introduction will give the reader a better perspective on my biases and on how to interpret my review. Every listener gives a different weight to different parts of the reproduction. That is why I separate my review into different sections: timber, soundstage, dynamics and transparency.
Timbre & Tonal Balance:
With the dac19dsp/c2 combination, it is in fact difficult to describe the sound through the traditional bass/midrange/highs as it would implicitly imply that the sounds of instruments and voices are disjointed and that it is always the same regardless of the recording.
What the dac19dsp/c2 combination achieves is a sound that is whole which is a little bit harder to describe. No part of the frequency spectrum seems to be calling attention to itself.
So how does it sound?
Let’s start with the tonal balance. When using the ACSS connection, it is dead neutral. Switching to the RCA, the sound was mainly dependant on the opamp being used on the C2 headphone amp. I tried the stock OPA2604, Opa Earth and Opa Moon. While the Opa Earth seemed the most “neutral”, I ended subjectively preferring the OPA Moon when using the RCA. I also have to admit that I didn’t listen much to the RCA inputs as it was totally outperformed by the ACSS input.
So in order to simplify things, from this point on, and unless I make a direct reference to the RCA connection, I will be mainly describing the sound of the system through the ACSS connection.
The most striking quality of the combination is its richness of tone. It is unlike what you can hear with certain tube amps where everything sounds pleasing but rather similar. With the DAC19 DSP and C2, there is simply a finer gradation of the sounds.
If I were to make a video analogy to make a comparison, I would say that listening to the Audio-gd FUN for example is like setting your PC monitor to 16 bits, while listening through the DAC19 DSP & C2 is like setting your monitor to 32 bits.
But that would only tell half the story. If we stayed in the visual analogy, the DAC19DSP+C2 would also have a higher contrast ratio (with deeper blacks and brighter whites) as well as more faithful colour tuning.
Staying with the visual analogy, I have friends who like to set the contrast ratio of their TV a little bit higher than neutral to increase the feeling of “high-definition”. While that might help with some particular situations, it kills the finer details and everything starts looking overdone and the same.
The DAC19DSP/C2 is like a finely tuned TV. It doesn’t grab your attention at first sight but it is only after watching it for a while, with different sources that you start realizing how good it is.
That is to say that the dac19dsp/C2 combination doesn’t have the pleasing euphonic coloration of some tube components nor does it have that overly sharp presentation of some solid state gear that could be associated with a high definition sound.
Personally, the relative neutrality of the dac19dsp/c2 is exactly what I was looking after but I understand it might not be what everyone is consciously or unconsciously looking after in an audio system.
So while the DAC19DSP/C2 is a dead neutral association (as far as I could tell), it doesn’t mean that it is sterile sounding. It won’t add anything to the recording but it won’t subtract anything neither. In my system it gave me the most tonally neutral and rich sound I have had the opportunity to listen to.
I had previously described the dac19mk3 (with the PMD100) as analog sounding. The dac19dsp goes a little beyond that having a more realistic timber. It is a like if the new DSP-1 filter not only combined the strengths I have identified with the PMD100 (analog sounding, accurate timber) and the DF1704 (fast and analytical), but it also added a little bit more realism to the mix.
One of the first albums I played was Natalie Dessay’s Italian Operas album to assess how good the dac19/c2 was. That opera was recorded at very low volume with some complex instruments and a big dynamic range. With lower end equipment, that CD can sound pretty horrible.
With the dac19/c2 combination, the voice of Natalie Dessay was superbly defined (on lesser systems it can turn into acid), and it was so realistic sounding that you get the feeling of having a living, breathing person in the stage in front of you. The accompanying performers were also beautifully rendered and differentiated. But the thing that I was looking to check was how good the dac19/c2 rendered the glass harmonica which is extremely difficult to reproduce convincingly. With lesser DACs (or even amps), the glass harmonica sounds like a bunch of weird annoying noises. The only DAC that made the glass harmonica sound right (and not an awful noise) was the dac19mk3 with the PMD100 HDCD digital filter. So here the dac19/c2 did a pretty good job but giving a more accurate and detailed representation than my former reference system.
One of the favourite tracks that I like to play lately to evaluate new equipment is the “Use me” track from Patricia Barber’s Companion Album. The solo bass playing at the beginning of the track is a good indicator of bass extension and articulation, as well as the richness of tone. On lesser equipment, the bass can sound light, dry or bloated. Here, the dac19/c2 does an excellent job by rendering the bass with excellent articulation. You can hear distinctly the vibration of the strings as well as the vibration of the body. And while the bass is very deep, it is very articulate at the same time and doesn’t obscure the low level details. Throughout the track and the album, the sounds of the instruments don’t mask the ambient cues (people chatting, glass noises...) that let you know the recording was done in a live jazz club environment. You just feel being there.
When listening to “A Case of You” from Diana Krall Live in Paris Album, the piano is beautifully rendered and you get all the little details like the multiple coughs and the small whisper/exchange around the 35th second. However, the most striking quality of this track is the way the dac19/c2 renders the way Diana Krall is singing. You get the feeling of a real person in front of you singing, articulating each note. You can almost “see” here face changing expressions while singing. So the dac19/c2 not only renders the fine little details, but it is also able to render the emotion behind what is being played.
Glenn Gould – The Goldberg variations: The piano is full bodied and firmly grounded; the weight of the presentation is preserved unlike some other experiences where the piano seemed to float around.
There is an excellent rendering and distinction between the attacks, sustain and decay of the notes in this recording.
Up until now, the best representation I had with this particular recording was with the dac19mk3 with the PMD100 digital filter; however the attack of the notes were a little bit blurred. On the other hand, when I used to switch back to the DF1704 filter, the representation would get clearer with more distinct attacks, but at the same time it would get a little bit “glassier” sounding than the PMD100 and the notes would decay too quickly.
Keith Jarrett – Paris London Testament: Here is another beautiful piano recording. While I prefer the Glenn Gould recording from a subjective point of view, the Keith Jarrett recording is better from a technical/recording point of view. I downloaded the 24/96 version from HD Tracks and it is true bliss to the ears. It is simply the most beautiful piano recording I have on my collection. It is breathtaking of realism and beauty. Here the DAC19/C2 reveals its excellent analyzing ability: you clearly hear the hammer hit the cord, and then you can distinctly follow the vibration of the strings being hammered as well as the resonance of the body of the instrument. Then you can follow the notes hanging in the air and slowly decaying.
The DAC19/C2 walks a fine line between revealing the finest inner details of each note while blending everything into one whole representation.
This recording also shows that he dac19/c2 has a clean, undistorted and wide frequency bandwidth. From the lowest to the highest note, you don’t feel that the dac19/c2 is adding or subtracting anything. The highs are sparkly without being glassy while the bass is deep and articulate.
Soundrama - The Pulse test CD: I usually use track7 (beach waves) to check the tonal balance of new equipment. When playing that track through the dac19/c2, it was the first time I felt I was at the sea shore, and that the waves were close by. When I closed my eyes, the only difference from the real thing is that I didn’t get splashed by the waves hitting the shore. That track was pretty convincing and as close as I had gotten to the real thing listening to that track.
What it means however is that the dac19/c2 has a very even and well extended frequency response from the very deepest lower notes to the highest ones. I had previously used that track (which is recorded at an extremely low volume) to highlight differences between interconnects and DACs and I have always found that it was easier to pick up differences through that track than through listening to music when comparing close sounding equipment as any flaw in the tonal balance is highlighted with this track.
Besides playing a selection of high quality files to gauge the performance of new equipment, I also like to play some low bitrate and highly compressed mp3 files. In fact, the effect of a low bitrate mp3 should be subtractive and not additive. With certain equipment that sound good on with high resolution material, they can sound pretty horrible with MP3s. The reason is not that they are revealing about the (poor) quality of those files but rather because they overemphasizing (minor) flaws about the encoding. In that regard, my entry level DACs using sigma delta dac chips (emu 0404 usb, FUN, Purepiper) are worse at handling MP3s than the DAC19mk3 was or than the dac19dsp is.
While the difference is pretty audible between MP3s, lossless 16/44 and 24/96 through the dac19dsp, all of them are enjoyable to listen to. My guess is that the whole PCM1704 uk, discrete design and zero negative feedback minimizes distortions that are overemphasized by sigma-delta, opamp based and high feedback designs.
Soundstage & Imaging:
While the soundstaging and imaging capability are excellent, they are nonetheless very dependant on the recording and the source than anything else. When I tried the Teralink X2 for a while instead of the modded Hiface that I used for most of the review, the soundstage shrank in size and became flatter.
However, regardless of the sources, the soundstage provided by the dac19dsp/C2/HD-650 expanded well beyond my head and was consistently better than my old reference and the other gear I had on hand at the time of the review.
What you won’t get with the dac19/c2 is a soundstage that is always unnaturally big regardless of what is on the recording. There are some tube amps or sources that can give you that “big sound” all the time but at the expense of a blurring of the sound sources.
What you will get is a soundstage representation faithful to what has been recorded and also a “being there” feeling. Thanks to its excellent low level details retrieval, the dac19dsp/c2 preserve all the little ambient cues of the recorded event that fool you into thinking you are there instead of just listening to a recording. That is especially noticeable with well recorded material and most live recordings. In fact, that “being there” feeling is also apparent on watching movies. You just feel immerged into the movie which no speaker system (whether at a home or at thx movie theatre) has ever achieved to such a startling degree.
On of the strengths of the dac19/c2 is to play very intricate and close sounds while keeping at the same time everything distinct and with respecting their true tonal colours. Some components achieve “razor sharp” imaging at the expense of the tonal density and diversity. This is not the case here where you get both pinpoint and full bodied imaging.
Another strength that seemed to emerge from playing different well recorded albums with the dac19c/c2 is the “stability” of the soundstage (that I explain in more details in the musical examples below).
I was also surprised by the phenomenal depth of the combination, which was more apparent with the ACSS connection by the way.
However, that big soundstage and especially the depth was pretty disturbing at first. In fact, I got used to sounds getting slightly pushed into the foreground in loud passages. The dac19dsp/c2 combination does it differently and everything stays at its place. This gave me at the beginning the impression of a slight deficit in impact. But after listening to it after a few hours, I don’t think I will ever be able to adjust back to the old way that sounded more “compressed” to my ears.
Since the whole soundstaging through headphones is very subjective, I thought it was necessary to mention it as some people might be looking for a “tighter” and more “impactful” representation. While that is not the case through the ACSS connection, you can achieve such a result through the RCA input (depending on the choice of the cable).
The first time I played the 24/96 Marianne Thorsen – Mozart Violin Concertos through the dac19dsp/c2 I thought for a few seconds I played the wrong file. The sound was so “big” that I honestly didn’t recognize it at the beginning. And I had listened to that album countless of times.
Imaging wise, I used to be able to focus on the soloist and “see” her perform. But the rest of the performers were diffuse. With the dac19dsp/c2 it is like everything got cleaned up and I could see clearer in the farthest recess of the soundstage. You can almost count the number of performers.
While the most striking aspect of this album played through the dac19mk3 (with the DF1704 filter) was the holographic imaging, now the most striking feature is the excellent separation of instruments and the fine capture of the tonal characteristics of very close sounding instruments.
When playing Mahler’s Fifth symphony, it is easy to get a cluttered representation with entry level gear. Here thanks to the exact rendering of timber of instruments and the solidity of the soundstage, you just feel like being there and listening to a real performance. The instruments stay locked at their place and do not rush in the foreground. The sense of depth is amazing in this recording and you get that the instruments in the background are really far away. You get an excellent depth layering and can distinguish pretty easily between instruments playing at a low volume in the foreground and instruments playing at high volume in the background.
Playing the 24/88 Dunedin Consort – Messiah, the dac19/c2 excels at rendering a holographic imaging of the scene. Every performer is again firmly grounded at his place in space. There is a better distinction between various performers and the accompanying instruments. It seems like there is more air and space between the performers in comparison with my older reference.
While listening to the track 5 of Soundrama – The Pulse, which is a recording done in an outside mountain environment, you not only get a big headstage, but you also get the feeling that the space extends indefinitely outside of the head.
The dac19/C2 combination has an effortless rendering micro-dynamics (small inflexions) or macro-dynamics (big changes). Whether it is big orchestral explosions or small inflexions in voices or a violin, the dac19/c2 does it with ease, refinement and effortlessness.
The dac19/C2 combination can be played at low volumes or high volumes while retaining exceptional dynamics capabilities. The C2 drives the hd-650 with ease and effortlessness. They make it seem like the hd-650 is a high sensitivity headphone. It has an excellent control over the hd-650 but without over-damping them. From memory, I would say that the older C2C had more grip over the hd-650 but I am not too sure about it as I didn’t try them side by side.
While I have been very impressed by its dynamics, I felt that the transients were somehow softened though the ACSS connection. I am not sure if it is due to the lack of perceived distortion, to the burn-in (I am still in the early stages of burn-in) or to the ACSS cable. Back, when I compared the Sharkwire RCA to the excellent Artisan Ultimate Silver Dream, I sensed it lacked speed and that it blurred a little bit the transients. However, since I had only the stock Shark ACSS cable that Audio-gd sent me, I couldn’t test that theory. Switching back to the RCA and the Artisan Silver Cable, I got back that transient snap and speed that was relatively missing from the ACSS but it was a step back in every other area.
Dark Knight OST: I have to admit that I have never listened to this track with this level of clarity. It was slightly congested and cluttered with my previous reference system: I use to have those big swings in volume but I thought that it wasn’t recorded perfectly. I just realized now that the quality of the track is pretty amazing.
When listening to “Aggressive Expansion”, the speed of the system is amazing. Even when I know that there is a big dynamic swing to expect at the 17th second, I still get surprised when it “hits”.
While my previous reference system was good in that regard, it feels compressed now in retrospect. As for making comparisons with the other gear I have on hand (Purepiper, FUN, emu 0404 usb...) there is simply none to be made. All the other equipment I have either turn to acid (for the DACs) or just give up (for the amps) at such high volumes.
It is amazing how the Sennheiser hd-650 can scale so well with better gear.
The system can throw a “big” sound which doesn’t make me wish I was listening through big tower speakers.
Gladiator OST – Battle track: There is no hint of compression. The representation is even slightly more dynamic than the dac19mk3 and C2C which were no slouch in that department. The stability of the soundstage is the same regardless of the intensity of the passage. However, the most striking feature of the dac19/c2 listening to this track is how the tonal beauty of the instruments is preserved even in loud passages. Here the dynamics do not come at the expense of tonal accuracy.
Puccini - La Bohème: It is a relatively old recording but that is an excellent test CD for judging dynamics. The album alternates a series of very quiet passages and very loud ones: the microdynamics are very well preserved and every little intonation of Pavarotti and the accompanying performers are very well rendered the way you would expect a good tube component would do it. However, the dac19dsp/c2 has no inertia and can follow the pace pretty quickly. Most importantly, the dac19/c2 preserves the flow of the music.
Overall, the overall performance brings back to life this old performance.
Transparency & definition:
If you had read the previous sections, you probably would have guessed by now that the dac19 and c2 are very transparent units.
I have noticed that some people confuse sometime transparency with mid treble brightness. In my opinion, that mid treble brightness and forwardness exhibited by some “pro” oriented gear such as the EMU 0404 usb is not a sign of transparency but another sort of colouration.
Personally, when I try to evaluate the detail retrieval of components, I try to not concentrate as much on the detail that jump at you at first but more on the low level stuff. In fact, most decently built DACs and amps will render correctly the most audible details. The differences we hear between them have more to do with the voicing and how the details are rendered in the soundstage.
What higher end gear get you is a better retrieval of low level details and the dac19/c2 are terrific at digging very deep on the recording and extracting every last bit of it.
While the overall resolution of the dac19/c2 is superb, it can be best described as “relaxed high-definition”. All the details are there, you don’t need to concentrate to hear them but at the same time those extra details do not distract you from the main performance but at the contrary help getting a more realistic one. It is what gives that “being there” feeling that I described on the previous sections.
Also, you can listen for hours to the system without having any listener fatigue.
I have seen a lot of comparisons/reviews where people say that component A is more detailed than component B. Most of the time, that assertion is biased by the fact that A is brighter than B. With the dac19/c2 that resolution doesn’t jump at you, but after listening a few minutes, you discover a new way and more realistic way to listen to music.
There is also one more interesting thing to tell about the resolution of the dac19/c2: whether played at extremely low volume level or a very high one, the dac19/c2 has an excellent rendering of the fine details. I think it is one of the qualities of the ACSS volume control as I had previously noticed the same thing with the C2-C amp. After listening for a few months to the older C2-C amp and now to the C2 amp, I find that more conventional methods of controlling the volume loose too much resolution at the lower volume settings.
Note about high resolution audio: Before the dac19dsp+C2 I have always been able to distinguish mp3s from lossless sources (I did a few ABX tests in foobar to confirm that). However, I had found that while we can hear big differences between 44.1K and 96K, it is mostly due to the poor digital filtering at 44.1K. That is why I have always felt that 16/44 up sampled to 24/96 gave relatively close results to native 24/96. This was the first time I have clearly identified 24/96 files as clearly superior to 16/44 not just different. The biggest differences I noticed were better micro-dynamics, more air around instruments and voices, a more accurate and richer timber. Overall, it felt less like reproduced music and more like the real thing. Given that the combination was already doing an exceptional job at 16/44 it is just amazing how reproduced audio can get.
Needless to say that the dac19dsp+C2 combination will only be as good as the recordings it is being fed. It doesn’t mean it will disappoint with low bit rate mp3 as it will in fact do a better job at not overemphasizing their shortcomings. But when the combination is fed with high quality files or high resolution files, you get transported to another place.
The transparency of the dac19/c2 was also apparent when trying different headphone cables. At the time of the review, I had on hand 3 aftermarket cables for the Sennheiser HD-650. The 3 of them are pure solid core silver with a few variations in the dielectric (and a cryogenic treatment for the Revelation Audio Lab one). While I had a hard time distinguishing significant differences between the cables when using the entry level Purepiper DAC A-1 and Audio-gd FUN, the sonic characters of each one became more easily audible with the dac19/c2.
I have also mentioned earlier that the dac19dsp/c2 lacked a little bit of transient snap but to be more exact, it was the case with the Artisan Silver Cables Mark1 (and of course the stock cable). When I switched to the Mark2 Artisan Silver Cables and the Cryo treated Revelation Audio Labs later that speed was back.
Note on ACSS vs RCA:
While I had previously listened to audio-gear that incorporated the ACSS technology, there was now way to tell how much of the “good sound” was due to the current gain and how much was due to the rest of the circuitry.
It was until I tried the Audio-gd FUN that I had the opportunity to compare the current gain technology vs. the voltage gain technology on the same component. I was pretty much amazed by difference there was between the voltage gain module (which uses voltage feedback I assume) and the current gain module (zero feedback). For more details, you can read my review of the FUN here: http://www.head-fi.org/forums/f5/rev...preamp-468522/
So before getting the dac19dsp/c2 combination, I was planning to compare both methods of connections using the same Sharkwire interconnects to keep the comparison fair...or so I thought. I was supposing at the time that even if the ACSS connection was technologically superior, there was simply no way that a “cheap” $20 ACSS wire would outperform my 15x more expensive reference RCA interconnects (pure solid core silver, air/Teflon dielectric).
So I was very surprised (to say the least) when I found out that the basic ACSS interconnect provided by Audio-gd totally outperformed the Artisan Ultimate Silver Dream RCA.
The improvements the ACSS connection brought over the RCA was a removal of electronic grunge present on the RCA connection, and it also provided a more spacious and more natural sound. The weird thing is that I was unaware of those faults before. But now that I have listened to a purer path, it will be hard to go back. Given that the C2 amp + a pair of ACSS cables cost not much more than a pair of high end interconnects... this makes it very difficult to beat.
Measurements (ACSS vs. RCA):
While I did a few measurements with RMAA to check a few parameters, I decided however to not copy past the results because most people do not know how to interpret them.
Indeed, RMAA captures only a partial picture of the performance of components as it measures only the frequency domain performance and leaves the time domain performance unattended. While many manufacturers focused their attention on the frequency domain of their gear, many of them left aside the time domain performance to which the human ear is very sensitive (you can read these research papers for example: Information for prospective students).
What I was curious to measure were the differences between the ACSS and RCA and the way the dac19/c2 handled the volume control.
I did my measurement with the EMU 0404usb, and measured the dac19dsp+c2 with both the ACSS and RCA connections.
I measured both connections with a SNR of 106db. However, you have to keep in mind that the RCA input on the C2 had 3db more gain than the ACSS.
When I lowered the volume control a few db, there was no change in the SNR until I start hitting the noise floor of the emu 0404 usb.
This reminded me of a measurement Stereophile did on the $18,500 Ayre KX-R preamp that also had an unconventional gain volume control.
That seems to indicate that thanks to its variable gain control (in the current domain), the dac19/c2 is able to retain its full SNR over different volume settings (which correlates with what I heard in my subjective listening). Usually by lowering the volume by 3db for example from the best setting, you get a drop in SNR of about 3db. That wasn’t the case with the dac19/c2.
Usually by chaining a DAC and an amp for RMAA, you can worse results than when measuring a single component (because of different gain settings, cumulative distortions...). Here I found that the dac19/c2 had a relatively wide bandwidth (it was down only by 1db at 40khz) and it had a low 0.008% THD for 2 components using zero negative feedback. In fact, it is easy to achieve lower level of distortions (the emu 0404 usb has a level of 0.0007%) but it is usually done by heavy use of negative feedback which usually cause time intermodulation distortions that a lot less ear friendly than the THD. So it is a sign of an extraordinarily good design to achieve such low levels of frequency domain distortions with a zero negative feedback design. If you are curious to see how other zero feedback designs fare in classical measurements, you can look at Stereophile archives, and you will probably surprised of how high the THD can get high in those devices without being problematic for the listening experience.
List of bugs encountered:
During the few days I have been using the dac19/c2, I noticed a few quirks that I thought worth mentioning.
I have noticed that there is a channel leakage in the C-2 headphone between inputs RCA 1 and 2: So if you are planning to A/B 2 sources at the same time from those inputs, there will be channel leakage.
At one point I was trying different upsampling rates while playing back. And after that I noticed that there was an inversion of channels in the DSP-1. After shutting down the DAC19dsp and restarting it, there was no channel inversion and I couldn’t reproduce the bug.
The ACSS connectors look a little bit cheap and are not very smooth to handle. I looked around for the price of ACSS/CAST connectors that are used by Krell and found out that the cost of the parts alone was $200 (4 connectors for the dac/amp and 4 others for the cables).
Limitations of the review:
The first point is that at the time of the review, I didn’t have anything better to compare to.
The second point is the interaction of the dac/amp with the rest of the equipment. While some people believe it is sufficient to buy a good dac, headphone amp and headphones to have a good sound, I believe it is quite the contrary.
In my experience, the “little” tweaks such as a good transport, power filtration, power cords, and vibration control contribute to a very large extent (more than 50%) to the final outcome. So if someone is planning to use the dac19dsp or c2 in a “poor” environment, the results can be unpredictable.
That is one of the reasons in my opinion why some people get disappointed after they buy a very expensive DAC or pair of interconnects and don’t notice much difference in their system.
With that being said, it appears that the first reviews of the dac19 DF and DAC 19 DSP are very positive. Also, if you check the dac19mk3 thread, most buyers were pretty satisfied with it and I believe that all those who mentioned upgrading did replace it with a higher end model from Audio-gd.
Edit— More on the limitations:
Just before finishing the review, I received a PM from a fellow head-fier who knew I was working on this review (he owns an audio-gd FUN and considers getting a DAC19).
Among the suggestions he made were: using a “regular” transport and digital cable, using the stock power cable, using no power filtration and stock interconnects. What he wanted to know in other words was how the “stock” FUN performed against the “stock” DAC19/C2 combination.
While I understand the logic behind what he asked, I wasn’t able to re-write my review to address those issues. However, there is a reason why I haven’t done any of my reviews with “stock” cabling and accessories. The reason is pretty simple: once I have tried different digital cables (I had more than 10 of them at the time of the review), I could detect the sonic signature of each one of them. Choosing a “regular” digital cable would imply that I would be describing that specific digital cable rather than the DAC itself. For example I have 2 relatively good entry level digital cables: the Canare and the Belden. So should I use one or the other (as they have distinctive sonic signatures)? Or maybe if I picked one, I would be mentally adjusting my perception of the sound to guess what you would have been the sound with a more transparent cable such as the Oyaide. The same would be true for the transport, interconnects, power cords, power filter, footers, platform support... During the last months, I have carefully selected those accessories by choosing each time the ones that sounded the most transparent and neutral to my ears (in comparison with similar equipment). So by moving back to the “stock” conditions I would face 2 difficulties: The first one would be to choose “regular”/”stock” components and then I would have to familiarize myself with that new set-up.
So when reviewing equipment, I try to use the most transparent set-up I have in order to describe the intrinsic character of the new equipment and not an aggregate of the sound of the reviewed component combined with very coloured associated equipment. The end result is that while my review won’t describe accurately what everybody will hear in their system, it is my best approximation on how the reviewed equipment really sounds.
Nonetheless I have to admit that his point was valid in a way that I didn’t assess more thoroughly the impact of those specific tweaks on the sound of the dac19/c2 vs. the FUN for example. I will try to address that point later on a follow-up.
When acquiring or reviewing new equipment, I always like to try a lot of different and complex material to assess their limitation. Here it was quite the contrary. Everything I threw at the system was rendered beautifully. It wasn’t by means of glossing over the details or by cheating through some sort of euphonic coloration. It was quite the opposite, by means of presenting all the material faithfully and without adding or subtracting anything that I could detect.
So what did I gain from my previous reference system which was composed with the following: dac19mk3 + Artisan Ultimate Silver Dream + C2-C (with stepped attenuator)? Now the music makes more sense, with better flow. The instruments and voices are lot more realistic, more lifelike, with better separation. “Complex passages” are not so complex anymore. The dynamics are realistic (i.e. no discernable compression). The units are very resolving yet “musical” and analog sounding. The sonu
Very big and defined soundstage, holographic imaging
And the biggest compliment of all, it is very transparent: you listen more to the music and less to the equipment
But is the dac19dsp/c2 combination perfect? The answer is obviously no as there always better things out there.
I suspect that the bass could go even deeper with other equipment and that the soundstage could get bigger. But the things that the dac19dsp+c2 combination achieves in a beautiful way is to do enough things well (and nothing wrong that I can detect) that it fools my brain into thinking that is realistic sounding.
While the DSP-1 seemed at first less sensitive to the transport, the dac19dsp didn’t reach its peak performance in my system until I paired it with jkeny’s modified Hiface. In that regard, the stock Hiface is a strict minimum to get a good performance from the dac19dsp, and the modded Hiface is highly recommended. The thing is that it sounds better than anything I had even with a poor transport, but it would be really a waste to not use a good transport with it.
But so far, the dac19 combined with the C2 have totally exceeded my expectations.
What is next to come:
Well, the improvement brought by jkeny’s modded Hiface was big enough that I am sending him my stock Hiface to apply the mods. For those intersted on Jkeny's modified hiface, I posted the review here: http://www.head-fi.org/forums/f7/rev...hiface-483900/
During the review, I used the stock Sharkwire ACSS that Audio-gd is selling. The ACSS connection is supposed to minimize the effect of the quality of the cable but not eliminate it entirely. So I am going to see if a different ACSS cable makes any difference at all.
I have already been in contact with Black Stuart from Head-fi to make me some DIY ACSS cables (He is also going to send me some of his RCA interconnects to compare to the Artisan Silver Cables).
I have also been meaning to try different oversampling settings on the DSP-1 and especially try the NOS mode for a longer period (I only tried for a few minutes to form a definitive opinion on it).
Once I have done those few things in the next coming weeks I will probably write a follow-up.
Edited by slim.a - 5/28/10 at 12:10am