Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Torq, Mar 1, 2017.
Labeling a component 'musical' is like telling your buddy his girlfriend has a 'great personality'.
As someone who lives in EU, the Soekris dac1541 is cheaper than the Gungnir Multibit. Difference isn’t big though, around $50-100. The Gungnir Multibit always seem to be out of stock in EU too. Probably something that can be addressed if I’d contact them, but I’m still considering purchasing the Soekris dac1541 to use purely as a DAC, and pair it with an iFi Pro iCan. You did mention this combination briefly in your review @Torq , but wonder if you found anything negative about this pairing?
No, nothing negative ... performance wise.
While I think the value proposition of the dac1541 gets skewed away from being favorable if you're not using it as an integrated unit, if it's cheaper, locally, than Gungnir MB then that doesn't really apply, I suppose. It'd still be nice to see a "DAC only" version of the dac1541 for such cases though, as that should be cheaper still.
For me, here in the US, and already having the Pro iCAN (and other amplification options) I'd almost certainly go with the Gungnir MB as it's meaningfully cheaper, has a signature I find preferable*, and I'd never use the headphone amp, switchable filters or digital volume control on the dac1541.
*Using Tube or Tube+ mode on the Pro iCAN provides a way to shift the presentation of the Soekris away from it's analytical leaning, somewhat. Which might be desirable, depending on your preferences.
I've decided to sell my dac1541, as I find it a touch too lean sounding with my active speakers. I find this perplexing, since it works so well with AKG K702s which in my opinion are certainly not recessed and in some sense even brighter than the Genelecs.
@Torq , any recommendations what to try next? I reckon it needs to be something with less upper midrange / lower treble. I've been thinking about Schiit Gungnir MB + Jotunheim or maybe a Holo Audio Spring DAC Level 1 with something that can act as a headphone amplifier and as a preamp.
Still curious is it would be so terrible to just EQ it.
I kinda agree with this when talking about low/mid end components. With some of the better high end gear though, that can mean no brightness or glare to the signature without having to sacrificing the resolution & detail.
Not really an option. I listen mostly via my computer but I listen to CDs too ( with an external CD transport ). Equing would require an external DSP processor.
I'm not sure if you're talking about my comment re: balanced vs. single-ended or about the use of the term "musical", so I'm assuming the former ..
I would challenge the designation of gear being "better" or "high end" (or even "better high-end") if it exhibits brightness or glare regardless of whether it was singled-ended or balanced (or, for that matter, whether it was "analytical" or "musical"). Unless "better" or "high end" is simply a reference to price.
Gungnir MB, Spring DAC, PS Audio Stellar Gain Cell DAC (not heard it, ESS-based*), maybe the Amethyst (not heard it).
The PS Audio unit might be a bit on the bright/hyper-detailed side of things (purely speculative based on my history with ESS-based designs), but it's closer to the overall functionality of the Soekris unit than the other three on that list (i.e. has a built-in headphone amp, digital AND analog inputs, pre-amp output, selectable filters, remote control).
I had an Audiolab 8200CDQ with an ESS9018 dac chip once - and I still have nightmares about the treble. So no thanks to the PS Audio unit.
Sorry to bother you with this, but how would you order Yggdrasil, Gungnir MB, Spring DAC and Soekris in terms of upper midrange/lower treble forwardness?
Before I do that, you get to take a stab at what you think I'm going to say based on what I've posted about these units over time.
My best estimate based on your writings ( from most to least forward upper mids / lower treble ):
Soekris > Yggdrasil > Spring Dac > Gungnir MB
However, if this is the correct order, I have no idea how big these differences are between each consecutive unit. The change or difference I am craving here for is certainly not big.
Pretty close ... from most to least "forward" upper mids/lower treble, I'd put them as:
dac1541 > Yggdrasil ~= Gungnir MB > Spring DAC
The tonal differences in these regions are quite small between these units. You'll find bigger differences at the frequency extremes; for example, most of the tonal difference between Gungnir MB and Yggdrasil is found below a couple of hundred Hz or so). And the Spring DAC will be rather less forward in the extreme treble.
What tends to happen, perceptually, is that you'll find some aspects of the presentation exaggerated based on differences elsewhere in the spectrum. In reality, there's probably no difference in the actual "forwardness" of, say, the dac1541 and Yggdrasil in these regions ... but you perceive a leaner sound, and the impression of more forward upper mids/lower treble from dac1541 due to the overall presentation of Yggdrasil carrying more weight/providing more foundation lower down. And the sweetness inherent to the Spring DAC tends to reduce the perception of forwardness.
Thanks for the detailed explanation!
I am now seriously looking at the Holo Audio Spring DAC. However, the engineer in me is seriously doubtful of anything that is designed to be NOS, but on the other hand, what the hell if it just sounds good. Something that is troubling me though is the characterizations about it sounding more in-your-face than yggdrasil, eg. first row (spring dac) vs fifth row ( yggdrasil ).
P.S. I have one suggestion to your audition list @Torq : Cambridge Audio 851D.
If you can't land a specific audition, at least try and do a back-to-back comparison between OS and NOS DACs.
Stage, with speakers, was more upfront, but also artificially wider than Yggdrasil. @Marvey found a rather different presentation of stage in his setup.
I'll add it in a bit.
When I'll get around to listening to it ... that's another matter, but it is, at least, pretty easy for me to get my ears on.
The Ares is a discrete R-2R multi-bit, oversampling, DAC with fully balanced and single-ended outputs, multiple digital inputs, support for DSD and a claimed* 20-bit resolution.
R-2R/multi-bit converters, especially those employing a discrete implementation are very much “en vogue” at the moment, and coupled with balanced output capability, enjoins the Ares as a member of a very small club - the more so given it’s comparatively modest asking price of about $650.
Thanks go to @alvin1118 (of Vinshine Audio) for making this unit available for measurement (by @atomicbob), and then for a full-on SBAF review tour.
This is a solidly made unit and feels significantly more substantial than you would be inclined to think just from pictures. This is not just down to heavy casework (though the front-panel is quite hefty); there’s a fairly beefy transformer contributing to the feel of the thing. A nice touch, on the exterior, is the application of isolation cones rather than simple feet, to the bottom of the case.
The overall finish is nicely done, the momentary switch-gear is firm and decisively “clicky” and the indicators are subtle and won’t light up your room at night. They remind me of Jony Ive’s approach to embedded LED indicators in Apple products ... they’re tiny and not obnoxious. This works for me as, in most cases, these indicators (input, sample rate, power/standby state) are something I use to occasionally verify state/function and not something I look at a lot. The indicators might, in fact, be a bit too subtle if you’re the kind of person that actually pays attention to them a lot, but as it stands I like them as they are.
Size wise, it's similar to Bifrost, but about half as deep again ... shown here in my rack for reference (I swore this didn't look that dusty when I took this picture ... guess I know what I'm doing next ...):
The Ares offers five inputs of three types, specifically one USB port and then four S/PDIF inputs via dual COAX and dual TOSLINK inputs. These are directly selectable from the front panel, allowing up to five devices to be connected to the Ares at once.
As always, one of the first things I do with any unit is to evaluate which input sounds the best to me so that I can do the bulk of my critical listening with the unit performing as best as it is likely to do. And in the case of the Ares it did not take long to settle on the coaxial S/PDIF input as the best way to drive the unit - which is how the majority of this review was performed. Driven via USB I found the sound to be unappealingly hard. Not just “harder” as has been the case with some units, but just, well, hard. Additionally, the background suffers a little and is somewhat less black than is the case when fed from a quality S/PDIF source.
S/PDIF interfacing duties were handled by both a Schiit Eitr and an Auralic Aries. I found no audibly discernible difference between the two.
Changes in this “Review”:
This particular review, or write-up, will differ a little from the others I’ve done as part of “Life after Yggdrasil”, and that’ll be through the inclusion of some commentary intended to map what I’m hearing to what I think that should resolve to in terms of measurements. This is an attempt to better to understand how what we hear is manifested in terms of objective performance.
I do not, generally, look at measurements before listening. It happens, from time to time, but it’s not my preferred way of doing things. And the kinds of measurements that would be interesting or even potentially useful there aren’t generally available anyway. In the case of the Ares and this review the measurements have been taken by @atomicbob and he is keeping those to himself until after I have posted my full write-up.
I am most interested to see if what I’m hearing (or, I guess, “think” I’m hearing ... though I’m pretty sure about that part of things at this point) is at all reflected in the measured technical performance of the unit at hand. Since the measurement results are not known to me at this time, all I can do in this post is speculate ... and proper discussion of those results necessarily has to be deferred until both sets of information (this review, and the raw measurements) are posted.
Hype and TL;DR;
I am aware that, on some “other” sites, the Ares has been billed, actually “hyped” is a better word, somewhat breathlessly, as an Yggdrasil and/or Spring DAC beater. And that it has been elevated to “giant killer” status in the process. This would be a remarkable feat, by any measure, particularly given the significant price differential between these units. It is, perhaps, unsurprising that elsewhere on those same sites the Ares is being proclaimed as inferior to the Audio-GD S19. Talk about extremes ...
It cannot, of course, be both - especially not outside the realm of personal preferences. But the range of claims there shows how dangerous hype can be, both for the vendor/manufacturer and for the potential customer. Neither party is well served here due to the potential for disappointment and resultant backlash.
As far as SBAF is concerned we can put the brakes on that hype-train right now. It didn’t take much listening to rank the Ares above the S19. It took even less to place it well below Yggdrasil and Gungnir MB. Rather more listening was required to decide upon where I think it really sits, overall.
First impressions were of a musically tilted DAC with a somewhat moister presentation, with a rounder delivery and a little more low-end presence than it’s immediate competition. From the upper midrange into the treble things can get a bit edgy. What this means to you in terms of desirability will depend on your personal preferences. It’s probably not quite the right DAC for “Captain Neutral”, nor is it the place to go if you’re pairing it with an amp or headphone with an uncouth top- end.
More protracted listening did, as it often does, reveal some artifacts that were not immediately apparent in short bursts of listening ... although now I’ve heard them I can’t un-hear them. The devil is in the details though, so let’s press on.
Listening and Sound:
Initial impressions were immediately of the notion that “this is different”. Not necessarily “wow” ... just, well, “different”. This has a tendency to form an initially positive view of things - since it takes you out of the ordinary. From an historical evaluation perspective, this can be a good thing and, sometimes, not such a good thing. This depends whether “different” is “better” or not. A prominent example of this is with the TotalDAC I auditioned last year.
Curiously I found that I preferred listening via the single-ended outputs rather than the balanced connections. This was true regardless of which amplifier I was feeding. I have some thoughts on why this might be the case further down. But the net result was that the overall presentation of the unit was smoother when using the single-ended outputs and listening fatigue was less of a factor (though still present).
As already mentioned I find this unit leans towards a musical, rather than an analytical, presentation. I’d stop short of describing it as “romantic”, but it is, for example, more likely to meet the preferences of someone that preferred Gungnir MB over Yggdrasil. Not to say that it is directly comparable to either ... both of those units show the Ares a clean pair of heels when it comes both to technicalities and portrayal of music, but in terms of the signature slant, that’s the sort of thing we’re looking at - or hearing ...
Compared to the Spring DAC, while perhaps tonally similar in the broad sense, it does not imbue the music with the same sense of “sweetness” that I hear with the Spring DAC. This isn’t terribly surprising, since that tends to be the result of NOS DAC behavior and the Ares is an oversampling unit. And while on the subject, while I don’t find any documentation to this effect, I am of the opinion that the Ares is upsampling all input to either 352.8 or 384 KHz (depending on the base frequency of the source material). I suspect this for two reasons ... first I don't hear any differences with higher- rate material vs. red book versions of the same thing (created from a common master file) and second ultra-fast electronic transients seem cleaner than I would expect for a less aggressively oversampling converter.
Dynamics are pretty good (macro, anyway ... micro-dynamics are about on par with a Modi MB or Bifrost MB, but do not match Gungnir MB). If I had to hazard a guess I'd say this is resolving about 16 actual bits, maybe not quite a full 17 bits of high-resolution content (or if so, not consistently for both channels) ... so effectively 16.5 or so (yes, I get that that's not strictly possible ... but it doesn't seem to be nailing 17 bits and it's an easier way to describe it for me).
Frequency extension is good on both ends (measurements for other DACs almost always show a neutral profile ... never seems to stop them sounding different though ... so I’m expecting that here), with a little more emphasis in the lower registers than I would consider strictly neutral. The top end exhibits good space and air, but there is some grain there that seems more prominent the higher up the spectrum we go. This has a tendency to make cymbals sound harsh and makes discordant brass exhibit more bite than is really there.
Listening to Julia Fordham there was an edge to her voice that, while not really sibilant (as often happens), was bordering on uncomfortable. But run some EDM or other more bass-centric music through the thing and it’s happy to dig in and party on. Perhaps the bass texture is a little less refined than with some units, but that could be down to my perception of it being slightly lifted.
My piano references, and pieces with plucked strings, showed a slight softening of the leading edge of the notes/attack, this is not uncommon for oversampling units. Still, I find this a bit less convincing than other, similarly targeted, units.
Many of Mary Black’s pieces I find very emotive - the intro to “Ellis Island” is almost always one that grabs me immediately - and if it doesn’t then there’s usually a problem. This emotional connection was easily made with the Ares, however it wasn’t something that I could hold on to. It worked initially, but over the course of several songs was something that just ebbed away and, ultimately, the music began to grate on me somewhat.
And in this ... the somewhat faster than I’m used to onset of fatigue or discomfort with the music, is where I find the weaknesses of the Ares, so let’s talk about that as it ’s important here.
The principle issue, from a “listening to music” perspective, I have found with the Ares is that, while initially quite engaging and musical in its delivery and able to put me “in the moment”, I can’t stay there for very long. I believe this is down to two things ... an increasing sense of edginess/grain as we progress up the frequency spectrum and the fact that, unfortunately, I am hearing differences between left and right channel performance that shouldn’t be there.
This is especially noticeable with low level passages/vocals, but either way one channel has audibly poorer micro-dynamic resolution, sounds more compressed, maybe even congested, exhibits more grain/edginess and has a rather more steely sense to it than the other. I think this apparent difference between channels is what is principally responsible for brining me “out of the moment”, and results in listening fatigue setting in comparatively quickly.
For example, I was reducing the volume before the end of Act 1 in “Carmen” and, while I didn’t nod off in the performance (which has happened with some units), I didn’t make it to the end of Act 3 either. This happens sometimes, and has happened with some very expensive/well-regarded units ... but it’s obviously not a desirable outcome.
I did not, as it happens, manage to pin down the specific differences between channels until I started playing some pieces (including some of my own piano recordings) that should be fundamentally the same in both channels and, well, clearly weren’t. Prior to that I was just progressively more aware that “something” was “off” and that I was becoming fatigued faster than usual.
This may be responsible for my preferring the unit via its single-ended outputs as well, as the difference was more readily audible via balanced. Which suggests that the balanced outputs might, in general, be the better way to go but, in this particular case, they’re making it easier to hear the differences between channels and the result of that is exacerbating the effect of those differences.
NOTE: This is one area where I am particularly interested in seeing if there’s an obvious correlation to measurements. In fact it’s the single most interesting point to me.
Digging In & Possible Resolutions:
Once I was aware of this issue I was able to conduct a few, somewhat basic, tests to see if they were specifically responsible for the fatigue and the sense of grain and compression. One of these involved using mono sources and splitting the output from the “better” channel on the DAC and using it to drive both left and right channels into my amp. Doing this resulted in a much more agreeable overall result.
Interestingly, if I swapped that to the “worse” channel, the result was still more agreeable than using mismatched channels. It didn’t take away the sense of compression or increased grain, but it didn’t result in the same level, or speed of onset of, listening fatigue either.
One of the bigger challenges with discrete R-2R DACs is in managing component differences and/ or matching performance between them or the modules they live in. With the Ares it looks like the ladders for both channels are all one board, so that makes it particularly difficult to adjust for any such deltas post-manufacture.
Moving the resistor ladders to discrete boards that could be swapped independently would make it relatively simple (if a bit more labor intensive) to more closely match channels and avoid this issue entirely. Of course, I have no idea if this level of channel delta is occurring in other units, or if this is just a particularly affected example (which would be unfortunate).’
Using my normal staple of native DSD material, some of which had to revert to using the USB input due to its bit-rate, I found that the Ares was a better performer than it was with PCM content. And by this I do not mean to say that DSD is better than PCM, simply that the principal issue I had with PCM content is not in evidence to the same degree when using DSD material.
Doing on-the-fly conversion of PCM to DSD also resulted in a better audible result than sending native PCM content to the Ares. Grain was reduced, there was a little less compression, and the onset of fatigue was significantly reduced.
At a guess I’d say that this is down to the being less of a delta between the DSD sides of the converter than in the PCM ladders (reasonable, since they’re smaller), but that’s PURE speculation. But if the performance of the Ares with PCM content was on the same level as it is with DSD then this would be a rather different review.
In terms of overall presentation I quite enjoy the fundamental rendering given by the Ares (I enjoyed it more before I honed in on the channel differences). It can communicate emotion and was able to draw me into the music quite readily ... it just didn’t keep me there.
I don’t think the Ares is a bad DAC - but I can’t overlook the effects of the differences I’m hearing between channels. It’d be a MUCH better unit for me if that wasn’t occurring (or wasn’t audible). That excepted, I could see, signature wise, that it’d be a decent alternative to a Modi MB or Bifrost MB if you wanted a multi-bit DAC and have a preference for a moister/rounder presentation and have a need for a DSD-capable balanced source. Though as stated I preferred this unit via its single-ended output.
While I think the Ares would pair well with amps/headphones that exhibit a drier/more technical presentation, I would be wary of pairing it with anything that exhibits bright, forward or grainy treble. The Ares is not bright in its presentation, but it does have a some audible grain that, for example, probably wouldn’t result in a synergistic pairing with something like Jotunheim.
If I was to attempt to place the Ares in terms of value/performance, it’d pan out something like this:
Gungnir MB > Soekris dac1541 > Bifrost MB > Modi MB >= Ares
There are, for sure, aspects of the Ares that might re-order that. Indeed, using a single channel doubled up would move the Ares at least one position to the left there, which is indicative that a better-matched output would improve my perception of its performance. And then output with DSD was notably better than with PCM content if that is important to you (this is not always the case). Specific system synergy would also have an effect given the less technical presentation of the Ares.
Another interesting aspect is what happens if I re-assess that ordering in the context of sending the Ares PCM content converted to DSD ... and in which case I get this (vs. the other units in a pure PCM chain):
Gungnir MB > Soekris dac1541 > Bifrost MB >= Ares > Modi MB
I find that interesting, if not necessarily a reasonable way to use the unit for most people.
At the end of the day, the inability to keep me immersed in the music, coupled with the faster onset of fatigue with the Ares means that it doesn’t really work for me overall. Perhaps so much focused listening exacerbates my sensitivity to the issues I’m hearing and maybe that’d be less of a factor for more direct, simple, “enjoying music” - it’s hard to say.
As it is, I’d like to see what the measurements have to say and then engage with @alvin1118 and see where we go from there.
*I say “claimed” as I have no doubt that there are actually the necessary 20 sets of resistors/resistor strings in the implementation. However, that doesn’t mean that, through the combined, cumulative, deviances from tolerance don’t result in an effective resolution that is somewhat lower than theory suggests.
Another great review that it's extremely well written @Torq. I was particulary interested in this one because of the amount of hype, even for hf standards. I wonder if the lack of balance between channels is something "easily" solved by adjusting the manufacturing process or not. Equally curious about @alvin1118 opinion on these impressions.
All in all, another great read and looking forward to measurements.
To be clear ... the difference between channels I'm hearing is not about "level" (volume), which is what "balance" implies to me.
It's possible that measurements will show that is also an issue, of course.
I would guess you'd need to move the ladders to separate boards so you could still build them en-massé and then just assemble the overall unit using matched boards.
It's possible there are other solutions too.
I was refering to the micro-resolution you referred to, maybe "balance" was poorly used in that sentence since it's most commonly used, as you said, regarding volume.
I have done a lot of thinking since writing up my thoughts on the Ares (actually some of this thinking resurfaced after posting my comments on the dac1541 as well). I'm conflicted, since, if I work past the issue with the channels being audibly different, it's not a bad DAC. Yes, it exhibits an increasing trend of grain up top, isn't the last word in micro-dynamics, and is perhaps a bit fuller in the bottom end than I'd expect (though that's not necessarily undesirable, depending on your preferences) ... but I could say the same about a lot of DACs, particularly with ESS based D/S DACs, some of which are significantly more expensive than the Ares.
I am, as ever, concerned that there are certain of us that look at DACs critically and that, when we make comments here about how something sounds, unduly or, perhaps, disproportionately, influences the perceptions and/or the interest of others in those products. Be that in a positive or negative direction.
Going forward I want to do things a bit differently and I am inclined to start that with the Denafrips Pontus.
First, I think, where possible, having @atomicbob measure units is very useful. Keeping those measurements to himself until the main "reviews" or "impressions" are published is also something we should keep doing going forward. Second, I think having several of our more experienced DAC-evaluators coordinate to publish their findings in parallel (not necessarily collaborating on those works, though that's an interesting potential), so that those independent opinions all surface at once.
Another thing to consider is having @Hands, @atomicbob, @Marvey (if he comes back) and myself at the END of each tour.
That last point can't be done for the Ares or Pontus, though we can of course hold back our thoughts until other's have heard those units. The only trick there is being able to get more participation in terms of posting meaningful impressions for those that are on the various tours for upcoming units.
And all of this would mean other changes to "Life after Yggdrasil" ("Life after Life after Yggdrasil" perhaps, but that might be getting a bit meta). That might not be a bad thing either - and would probably make it more obvious that it is no longer just Yggdrasil being used as a the reference point.
Just things to think about ... but right now I am leaning strongly towards not publishing my thoughts on the Pontus until others have had a chance to hear it and to comment themselves.
Separate names with a comma.