Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Maxx134, Jul 22, 2018.
Note: no Yggdrasil or Gungnir in the mix. Woe.... What will we do? How will we know what to think?
IDGAF what the SBAF hive mind thinks. I welcome fresh ears and opinions as long as they are respectful and try to add something to this wonderful hobby of ours.
A mega comparison here? Or individual impressions in each thread? I don't want @zerodeefex spanking me.
I for one want to hear your perceptions, particularly the Convert-2 and DA11. I've measured the former (posting soon I promise) and will be measuring the latter in the near future.
For what it's worth, I've been developing a database of measurements and constantly watch listening impressions in the (probably futile) attempt to find a correlation worth more rigorous investigation.
I think here is fine since it's DAC talk.
OK will do. I have about 10 pages of notes.. it'll take me a day or two to digest it into prose. If people have specific curiosities, please ask.
For which DAC do you have the fewest notes? I will infer that this is the DAC that deserves the most consideration because you were actually listening to music.
Take your time, obviously there is no actual rush on that but we all will be thankful.
The fewest notes goes to Hugo TT, cause it lasted six minutes before I ripped it from my system.
But the next fewest does happen to be the one I chose.
Well, if I may, and since you asked: I would be glad to read your impressions and I'm eagerly awaiting for it, so feel free to repeat what others have said if your findings are matching theirs. It's not like you aren't bringing anything useful because it's not all new and shiny, everyone has their personal perception, biases etc. so your testimony would add to the weight of those impressions. I'm pretty sure there will also be some new details that were more apparent to you, though.
In addition to the description of your findings as a text, I would find very useful if you would include a series of rankings such as (just an example):
Bass extension: Convert 2 >> Lavry >= Solaris > Metrum >>>PSA
The more criteria you would use for these rankings, the better. For example, I would be especially interested in comparing the dacs in terms of: HF extension, soundstage width, blackground, acoustic timbers reproduction.
Thank you in advance!
Having just finished a week-long DAC orgy with Dangerous Music Convert-2, Crane Song Solaris, Lavry Black DA-11, Metrum Adagio (Pavane + volume control), Benchmark DAC3, and PS Audio DirectStream (w/Snowmass), I will attempt to rise from this Satyricon stupor and report some personal impressions.
First, let me say thank you to SBAF and the loaner program for making this possible. Not only did I find a new DAC, I learned a huge amount in the process. Particularly about the role of jitter. Jitter had always seemed like a theoretical worry to me (like intersample overs) but now I’m convinced it is quite important. The upshot being that many of the criticisms I had initially noted when listening to DACs from USB turned out to disappear when listening from a high-quality source. Sometimes these changes were significant enough to reorder my DAC rankings.
Spoiler: More on Jitter
For those interested, I began learning about what jitter sounded like by going here. Those examples show that sometimes jitter can make things sound euphonic, like tubes; at other times crispy or fatiguing. I swapped around my three sources (MacMini USB, Schiit Eitr, and Allo DigiOne Signature) with different DACs to listen for different jitter effects. Such effects were clearly audible on all DACs when being fed from MacMini. Conversely, the theoretically best source (DigiOne) gave the driest and more accurate timbre on all DACs. Eitr tended to produce music that was more euphonic and warmer but also not accurate—except for on Adagio, where Eitr sharpened things up a touch but in a way that I did not like (like attempting to apply a sharpen filter to a low-res photo). In the end, I did not care for the Eitr sound. My personal choice was the DigiOne Signature for all DACs, which I presume is also the lowest jitter source of those I tried. FWIW, I cannot hear any differences with different power supplies on DigiOne Signature, though in all cases I kept wall warts far away from everything else. I could not detect (and should not be able to detect) differences between MoOde and Volumio.
I’ll try to give some quasi-objective measures of these DACs as well as my deeply subjective opinions. For the quasi-objective measures, I’ve noted tone, macrodynamics, detail, timbre, stage, imaging, transient response, and blackground. I’m never sure people use these terms in the way I do, so I put a little glossary of what I mean in the spoiler.
Spoiler: Glossary of Terms
Tone: The general equalization of bass, midrange, and treble. I may be biased towards a warm presentation based on what other people say, but I am not a fan of too much mid-bass creating a muddy sound. Tone is an interesting beast because there are obvious differences between DACs—some are warm, some bright—even when the frequency responses will all measure nearly flat. Differences in the measurements tend to be around 0.5dB (equal to ~12% in total power) top-to-bottom at best. This has led me to develop a theory (albeit unproven) that while the human ear doesn’t perceive “volume” changes to single tones at <1dB, the ear might nevertheless be capable of discerning tonal differences through the ratios of energy in each region. In other words, tonal perception might be closer to linear than logarithmic.
Macrodynamics: literally how dynamic (in volume changes) does the music sound as it progresses through a song. Does it sound compressed or exaggerated?
Detail: For me, this is closely related to microdynamics and so I roll them together here. I’m listening for warble in horns, and spittle jostling in the pipes. I’m listening for the bow of a cello snap as it reverses direction.
Timbre: This is an emergent property that describes how accurate individual instruments sound. It requires not only detail, but also the right spectrum of harmonic overtones. I make this assessment based on well-recorded live performances of acoustic instruments. Studio recordings sometimes use multiple microphones to give instruments a “magical” timbre that’s hard to calibrate. I find piano to be the hardest instrument to reproduce and thus a good test case. Timbre is for me one of the most important factors because I listen to acoustic jazz and classical. If you listen to pop/electronica, it’s almost irrelevant. Indeed, I’m starting to think many people prefer inaccurate timbre because they like the ways in which DACs can accentuate certain things, such as giving drums extra slam.
Stage: This is how wide the sound field appears, how 3-dimensional, and how far back the instruments seem to be placed. Emphasis in the presence region tends to bring the music closer in a more engaging manner, but it flattens the stage depth; deeper stage gives a sense of the room. You pick your poison.
Imaging: Distinct from stage, this is how pinpoint the instruments can be located on the virtual sound stage. I suspect (without much evidence) that big, wide, deep stage is largely the result of tonality shifts; but imaging requires accurate reproduction of the phase information in the treble region, which your brain uses to spatially locate sound.
Transients: Is a click a click, or did it become clack? Is wack becoming thwack (a sign of pre-ringing!)? Do cymbals splash and ring out, or just sizzle?
Blackground: When a whole lot is going on in the music, am I hearing a sort of echoy jumble, or do things sound clearly imaged against a quiet background. This is not the noise floor of the DAC (which can also be an issue) but a qualitative effect of reproduction.
Like I said, the above is how I use these terms. I’m not claiming to be the OED.
A note on method: I listened to these DACs on three different systems, one of which is a headphone rig. I feel this is necessary to really get a sense of the DACs independent of their potential system synergies and tonal balances. I never used DSP to attempt to correct for system tonality. Most test tracks are jazz, opera, or classical. I always throw in a few pieces of contemporary music, but these have much less influence on my views. Sources were USB, Eitr, and Allo DigiOne Signature—but ultimate decisions are all made based on DigiOne as the source EXCEPT for Convert-2, since I didn’t have an AES source I used USB with Schiit Pyst cable. Convert-2 was, however, always operated in the mode where it provided the master clock. Details of my rigs are in this spoiler:
Spoiler: My Systems
System One is Pass Labs XA30.8 driving B&W Matrix 801 Series III. These are the old B&W flagship monitors before B&W started making crap-sounding speakers. They are, however, bright, ported, with only modest bass extension. Bass is not tight. The Pass amp is incredibly transparent sounding, though it has the faintest hints of second and third-order distortion right at the threshold of hearing.
System Two is Ayre MX-R driving HHR Exotics TLS-II, a boutique remake of the old Walsh-F omnidirectional made with modern materials. This speaker is not flat at all. It has a lot of warmth and is rounded off in the high treble. However, it has no crossovers and is a kind of imaging monster. The Ayre amps are less transparent than the pass labs, but they have the damping factor required to drive the very heavy TLS-II.
System Three is my headphone rig: Pioneer M22 to HEK. Pioneer seems very uncolored to me and measures that way; almost dry. The HEK has a soft presentation and high treble issues. It does, however, reproduce microdetails better than the speaker systems.
Why no DSP? Because I think it does strange things to the music. Even the smallest little corrections or equalizations seem to screw up the imaging and timbre. On multiple occasions I’ve done experiments with small, judicious amounts of DSP (1-3 dB, f<<1Khz). But coming back a few days later, I will within minutes think something is wrong…. then I discover the DSP was left on. I don’t doubt that better DACs might still sound better than bad DACs, with or without DSP in the chain—but I’m of the view that after a certain price point, the best thing you can do to improve your sound is turn DSP off. And frankly, I don’t think I could evaluate a DAC properly with DSP on.
I’m trying to rank these according to some imaginary standard. I loaned Lavry out to a wayward DAC-less friend before I got the DigiOne source, so I feel I cannot give quasi-objective rankings for it. I’ll give some quick subjective impressions below. I sold off Benchmark DAC3 before I got DigiOne, but I feel I can rank some of its qualities against others based on memory.
Solaris > Adagio > Convert-2 > DirectStream > DAC3
Adagio > DirectStream > DAC3 > Solaris > Convert-2
Macrodynamics (ranked in magnitude w/ moar is better philosophy):
Convert-2 > Adagio > DirectStream > DAC3 > Solaris
Solaris > Convert-2 > Adagio > DirectStream > DAC3
Timbre (piano dominant rank):
Adagio > Solaris > Convert-2 > DAC3 > DirectStream
Stage depth (not width):
Adagio > Solaris > DirectStream > DAC3 > Convert-2
Solaris > Adagio (insufficient notes on the others)
Imaging (with DigiOne SPIDF source):
Solaris > Adagio > DirectStream > Convert-2 (USB)
Adagio > Solaris > Convert-2 > DirectStream
Adagio > Solaris > DAC3 > DirectStream ~ Convert-2
Note: Solaris has compressed macrodynamics relative to Adagio that makes it hard to compare blackground.
Purely Subjective Evaluations
Let’s start out with the surprise.
Dangerous Music Convert-2.
This DAC impresses right out the gate with whack-you-over-the-head macrodynamics. By the end of the week, however, I decided it was actually a novelty device. My love for it wore thin fast, and when I heard the other DACs from better sources, the comparative weaknesses of Convert-2 began to emerge. As Marv said, it’s listening to music in Vivid mode. I think it does that by (1) strong emphasis in the presence region, and (2) possibly artificial enhancement of macrodyanmics. THIS DAC IS YELLING AT YOU, and while it gets one’s adrenaline up in a positive way, it never stops. I also think there is a little tip up at the far ends of the spectrum, giving the impression of better tonal extension (sub-bass and >10k treble). I think this DAC could work well for headphone rigs where the emphasis is on being immersed in slam, detail overload, and intense engagement with music. But on 2ch, the lack of stage depth caused by the overemphasis in the presence region causes the stage to fall flat. Imaging is unstable and the instruments can hard-pan left or right, giving a U-shaped effect. Better suited for contemporary music with many synthetic instruments where stage is not an important feature. For acoustic jazz and classical, it doesn’t fare well. Quite apart from the brightness, it has no nuance, and rather bad timbre. Brush on snare sounds fizzy. Fundamental tones of most instruments ring clear, but lack proper overtones. A sense of wood or brass is missing. It’s a fun DAC. For many young SBAF headbangers, it’s probably a great match for their music and listening style. It’s not for old-fart “audiophiles” who drink cognac (like me).
And now for its dethroner.
Quite the opposite happened with Adagio. Initially I ranked it a middling DAC. Fed from USB, it had a polite, kind of warmed-over NOS feel—again a novelty sound. There was also lots of weird shit going on, like clicks in the music when played from straight USB. It sounded very warm, and positively muddy on my System Two. It was by far the most sensitive to source. Moving to DigiOne, all of Adagio’s flaws were remedied. It became crisp, the tonality more balanced, and it emerged at the end of two weeks as my favorite DAC. It renders with rich tonal density without aliasing and even-order euphonics. It is the undisputed victor on timbral accuracy, being the only DAC able to convincingly recreate piano. It has an uncanny ability to place instruments on a pitch-black stage. I wonder if this has something to do with the NOS architecture. It is the only DAC I tried that didn’t have low-level artifacts going on constantly in the background. One Soekris/GungnirMB friend found it spooky and actually seemed not to like it (too foreign sounding?). Here’s how I would compare the two: Convert-2 simplifies music by throwing away a lot of nuance and compensates by bringing out the pink highlighter and all caps. Whereas Adagio has a way of simplifying the listening experience by keeping so much of the music intact that it sounds simply natural. Of course, if you’re listening to Black Eyed Peas, Adagio is going to be boring, and Convert-2 will be exciting, so take that caveat very seriously. Adagio has one major limitation that will apply almost universally to all SBAF members: an insane price tag. If you’re a timbre fanatic like me, look for a used Pavane (Level 3) and get yourself a clean volume control for a more affordable mashup.
Crane Song Solaris
My experience with this DAC began with me proclaiming “yuck.” I wound up loving it and placing it a close second behind Adagio. It is the best-value DAC for people with my musical taste, but it has a particular signature that you must be willing to accept. Initially, when fed from USB (and especially having just listened to Convert-2) it had a compressed sound to it and a sort of wet cardboard timbre. The “velvet” sound of the AKM4490 chip doesn’t inject any adrenaline into your veins. Yet, there was also a brightness just at the very tip-top of the treble that made tape hiss a bear. Piling onto this, there was a nervous chihuahua hiding behind the velvet curtain—an ethereal, almost ultrasonic energy that I couldn’t quite put my finger on that made me want to give up on the DAC after about half an hour (whereas I wanted to give up on Convert-2 after 3 minutes). Switching from USB to the DigiOne, the chihuahua was shot dead and the velvet curtain began to feel more like flat lighting carefully illuminating every detail of the music without shadow or emphasis. I have to point out that the dynamics always remained compressed, but brain burn-in seems to compensate for this (at least for me) within a matter of minutes; and it becomes a feature when listening at night and at lower volumes. If I were a mastering engineer, I think I might choose Solaris because I could work with it all day and still hear everything in the music. By comparison, Benchmark DAC3 is also non-fatiguing, but it does so by killing off detail. With respect to timbre, everything sounds excellent (except Adagio still sounds a bit better). There is a very slight mid-range focus to the music, or maybe it is just that Solaris is not emphasized at both ends. Ultimately, it lacks the sweetness and blackground of Adagio, but wins out on ultimate detail.
Lavry Black DA-11
I have not heard this DAC fed from anything but Eitr, so I don’t feel I can judge it yet. However, from Eitr I felt the highs were lovely, clean, and extended. Very accurate, if not also a bit bright. I felt the bass was perhaps missing, maybe muddy (that’s Marv’s suggestion). Stage was narrow, but I liked it: very tight. It has a stage-width adjustment that actually works, but a friend of mine feels that using this feature to make the stage wider puts a hole in the center (I didn’t hear that). I actually enjoyed it with the stage narrowed further, because the imaging really begins to pop. It’s very affordable on the used market, and for treble heads, I might recommend it as a value option.
PS Audio DirectStream (Snowmass firmware).
My spouse says, “I pity the fool who buys a DirectStream.” But it’s actually our third-ranked DAC. Fed from USB, it has a terrible bloominess to the timbre and is extremely warm. It sounds more like what I expected NOS dacs to sound like. Juiced up creamy warmpoo would be an appropriate description. The detail and timbre were significantly improved when switching from USB to DigiOne, but it still falls behind Solaris. It also uses the trick of suppressing the macrodynamics a touch to make it very easy to listen to for long periods of time—though it is not nearly as compressed as the Solaris. DirectStream is actually easier to listen to than Adagio for this reason—you can jam out all day with DirectStream, but you couldn’t master with it. It also has pretty decent stage. The imaging is not nearly as tight as Solaris, but it’s not bad. Whether I would prefer Solaris or DirectStream matters partly on how important those dynamics were to me versus timbre and detail. I think for many people, macrodyanmics probably are more important, and so here DirectStream would win. That, to my mind, explains its popularity. One important note, this DAC has a bad noise floor. You can hear it hiss in a quiet room. The DS Junior is really terrible on the noise floor, and I would advise against the junior model.
This DAC falls into the category of tonally accurate, inoffensive, but lacking detail and imaging. I can’t say much more than that. I used a DAC3 for a long time because it was for a while the only delta-sigma DAC that I had found that didn’t bother me with etchy, crispy artifacts. But it has been been surpassed by almost all of the DACs above.
Queue me looking to demo a Pavane L3...
Cheers man, that was a fantastic read.
The L2 is identical save for 384kHz support, which you don't need, and it's considerably cheaper. Unfortunately you can only buy the L3 new now.
Time to warm up is Constant * 2^bits. Constant goes own with the age of the DAC. My 20+ year old Theta Gen V took a few minutes to warm up. This was side by side with Yggdrasil A1 and other DACs as reference, using the same tracks and same gear.
Ever since I’d owned (and sold) the Onyx, I’ve wanted to hear a Pavane, which I understand fixes a lot of the issues I had with the Onyx. Timbral density was just fantastic. But the prices are pretty high (although not necessarily compared to the PS Audio stuff).
@skem glad to see the DigiOne Sig helped you see some of these DACs for what they really were (rather than via their crappy USB input). Now you just need to get ahold of a Gungnir Multibit A2 or Yggdrasil A2 to compare. But based on your stated preferences, I’m thinking the Metrum may be more preferable for you, especially if timbre is really high on your list.
Nice observation that many audiophiles prefer worse sources due to the worse sound. I’ve noticed that too. Many like the perceived, muddying effects of jitter and unneeded reclocking of gear that for the least jitter possible should be synching to the clocking embedded in the original signal or have every piece of digital audio gear being slaved to the best clock in the chain, usually the modern converter or interface.
What rigs and recordings did you check the converters with? Asking as recordings vary a lot more than DACs and obviously anything with neutral treble and upper mids feeding B&W is gonna bring out the B&W laser beam etch or the diamond Death Star ray.
Expand the spoiler on Systems.
Recordings: Too many to list!
Good to know, thanks!
Man...Metrum + ATC actives could take me to Timbre Town
I definitely think jitter affects the sound & reducing it will give you better results.
Someone on Audiogon with an Adagio said he used Empirical Audio's Syncromesh, which reduces jitter, & it improved his DAC significantly.
There is also a newer & better jitter reducing device, albeit a little more expensive
I haven't committed to spending a grand on either device to test it out, but might in the future.
Fixing jitter has got to be much harder than avoiding it in the first place. That is the idea behind the DigiOne, as I understand it.
Separate names with a comma.