Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Hands]

First, Some Background…(And, yes, there is a TL;DR portion at the bottom)

You may have already read some of my thoughts on the Audial Model S DAC floating around on various sites or had a one-on-one conversation with me about it, but I wanted to consolidate my thoughts into a review of sorts. I especially wanted to write about it since it’s difficult to find many other people that have heard it. I went on quite the hunt to find this DAC and took a risk buying one. I’m sure it will be no surprise to some of you that this is a non-oversampling DAC, nor do I expect those that prefer an oversampled sound to change their mind with this DAC. After my experiences with various DACs, there’s just something about the non-oversampled sound that resonates with me, and my experiments with high resolution and upsampled content (with non or minimally ringing filters) has more or less confirmed that it isn’t just because of the inherent treble roll-off you get from NOS DACs with redbook content. As always, your mileage may vary.

I always liked the Metrum DAC sound, based on my experience with the Quad and Hex, but neither of those DACs were particularly resolving. They were also a bit too agile at times (I believe I used the word "sterile" back then), though the Hex much less so than the Quad. The Audio-GD NOS1704 DAC was noticeably more detailed than any of the Metrums, but the sound was too syrupy, the bass was a touch murky, it was a little grey sounding in terms of dynamics, the soundstage was small and lacked air, and despite its smooth nature, it had a slightly artificial, hard character on the top end.

Once I learned about the magic of some multi-bit, TOTL DACs from the 90s, I picked up a Classé Audio DAC-1. Even to this day and even when compared against something like the Theta DS Pro Gen. Va DAC, I still have yet to hear a DAC with as much bass slam and a three-dimensional, holographic soundstage as the Classé DAC-1 (either that or I just got used to those traits). It was incredibly lively and detailed. I’ve never heard a DAC that made my jaw drop as much as the DAC-1. Despite this and its fleshed-out sound, it was too fatiguing for me. Too lively up top, overly textured and detailed to the point where it didn’t sound as smooth as it should, and it was in general just very in-your-face. It had the sort of sound that immediately knocked your socks off but was relentless to the point where it became fatiguing.

While the Classé DAC-1 ultimately wasn’t for me, I finally started to hear and understand some of the magic these older DACs brought to the table, with their legendary converter chips or modules and excellent overall designs. I wanted to find a non-oversampling DAC that had strong dynamics, a good soundstage with plenty of air, didn’t gloss over details, and kept me emotionally involved. I wanted something that combined the strengths of what I had tried previously without too many compromises.

I tried a few different DACs and did a lot of research, for both subjective impressions regarding chips and certain fully-implemented DACs and objective data as well, before deciding to take the plunge on the Audial Model S. The TDA1541A chip has a relatively impressive set of specifications on its datasheet (not quite up there with the PCM63 or UltraAnalog modules, but still very good 16-bit with +/- 0.5 LSB typical or 1 LSB max, low distortion, etc.), and the Model S seemed to squeeze everything it could from the TDA1541A even with the difficulties presented by the non-oversampled design. My research on subjective impressions led me to believe I’d like the TDA1541A sound with the right implementation. Ultimately, the Model S turned out to be the right decision for me.

Some Quick Notes and Reference Points

First, I ended up getting the SPDIF version of the Model S. Pedja Rogic of Audial suggested it was a touch richer and less sharp sounding than the USB model. This is just relative, of course, and a very minor difference at that (i.e. both can be rich and sharp sounding). Though, it is worth mentioning that the USB model has inherently lower jitter than the SPDIF model. I was not particularly worried about that. You can read more about that on Audial’s blog portion of the site. I should also note quickly that the SPDIF version is limited to 16/96, and the USB version can go to 16/192.

My particular Model S includes the optional output transformers and RCA + XLR outputs.

I only had the JKSPDIF Mk3 and Gustard U12 USB->SPDIF converters on hand while gathering impressions. I do hope to pick up a better SPDIF converter in the near future. I did most of my comparison tests against the Theta Gen. Va, for reference (and I do quite like how the Gen. Va sounds), but even after installing a BNC input on the Theta, I could only get it to work with the Gustard U12. Part of me thinks the Theta may be less sensitive than the Model S to the input source, but, truthfully, I am just guessing and could be entirely wrong. Either way, I made sure to use the exact same setup when directly comparing DACs in an attempt to keep things as fair as possible.

As I understand it, Pedja Rogic is well known in the DIY community for his knowledge and experience working with the TDA1541A. His TOTL DAC designs seem to be quite robust, and I know that he uses a very specific batch of TDA1541A chips that he claims sound great and test very well. As I understand it, these are not re-used or refurbished chips either.

You can find the product page for the unit I bought here: http://www.audialonline.com/model-s/
This is the USB version: http://www.audialonline.com/model-s-usb/

I pulled this list of design highlights of the Model S DACs from Audial’s site:

● Advanced DEM clocking scheme for TDA1541A converter chip
● Novel hybrid output stage without feedback yet with top level linearity (0.002% THD)
● Options of transformers, capacitors or directly coupled output
● Isolating mains transformer with split secondary output (balanced AC power)
● Nine independent supply lines, each with three stage passive common mode filtering and zero feedback low noise regulators

As a side note, I just noticed that the MKIII version of this DAC was announced. It allows for simultaneous or I2S input mode without the USB implementation. It looks like there may be other slight internal upgrades, which would be impressive given how strongly the regular SPDIF MkII Model S performs already. You can read more about the MkIII here: http://www.audialonline.com/online-topics/model-s-mkiii-schedule/

Lastly, due to the type of architecture and elements used inside the TDA1541A chip, it draws more power and gets hotter than the sort of chips that came after. So, in order to prevent damage to the chip, it’s best to turn off the Model S when not in use. On the upside, it sounds its best after only a short warm-up period. I also noticed some improvement after letting the Model S play m usic for about 100 hours, leaving it off overnight. I’ve put a few hundred hours on it so far and don’t feel it’s going to improve further.

Model S Impressions

The very first thing I noticed about the Model S was that it seemed to impart less of its own traits in my music selections. While I had a pretty good idea of X album sounding pretty good, Y album sounding a bit grey and hazy, and Z album sounding a bit thin (or overly bassy), I find most DACs always overlay at least subtle, inherent characteristics to all music. Some DACs are almost imperceptible, and some are very noticeable. That’s not to say the Model S is entirely and absolutely transparent in all ways, but I was impressed with how honestly it seemed to portray different recording and mastering styles from different albums. Subtle mastering cues stood out nicely as well, such as one selection where the vocals are split into a left and right track. At the very end of the vocal line, the left and right vocal tracks quickly collapse into the center and come forward slightly. This was nicely handled by the Model S and stuck out to me in ways it had not before.

The Model S is also considerably more resolving than the Metrum Hex. It’s not even a comparison. Low-level details come through quite nicely on the Model S. Sure, you can’t expect the same type of sharpness, laser-like focus, and ability to dig up every atom of sound the best oversampling DACs provide, but when going back and forth between the Model S and Theta Gen. Va, very, very rarely do I find something that gets glossed over on the Model S. Some things might not be quite as clear or sharp, or sometimes things get a touch more buried, but I never find myself really wanting in these regards either when I switch back to the Model S. Really, it’s not even fair to lump the Model S in the same category as the Metrums or NOS1704 when it comes to detail retrieval.

Small things like the subtle plucks from finger picking don’t get lost. Someone shifting in their seat while playing piano or the subtle sound of the pedals and hammers moving up and down, also not lost (and I’m referring to recordings where the piano is not in the spotlight). You can not only hear those elements but sense the very subtle, somewhat low-pitched, soft thumps and air being pushed around from those elements as well. Along with this sense of air comes strong reverb and decay.

Like I said, it’s not quite as sharp, quite as focused, or quite as layered out as you’d get from the Theta Gen. Va or similar, but it’s close and certainly much closer than I ever expected from a non-oversampling DAC. At the very least, I’m happy it easily bests the NOS DACs I had heard prior.

I also noticed that moving to the JKSPDIF Mk3 greatly improved the sense of focus and sharpness on the Model S. It was NOT a subtle change. The Classé DAC-1, on the other hand, barely changed with different sources. I can’t say how sensitive to sources the Theta Gen. Va is in comparison to the Model S, unfortunately, but I touched on this a bit earlier. The Model S also gains a lot of ground with high resolution content. (As another FYI, most of my impressions were gathered while listening to redbook content.)

The Model S might also be the airiest, largest sounding non-oversampling DAC I’ve heard yet. In my experience, you can usually expect a smaller soundstage and less air when going the NOS route. Once again, yes, DACs like the Theta Gen. Va sound larger and airier with slightly better defined layers. I still think the Model S does very nicely here, NOS or not. On the other hand, while the Theta Gen. Va is wider, airier, and slightly better layered, it has more of a two-dimensional presentation to me. Each element of the music and each layer with these elements sounds flatter to me when considering a three-dimensional space. The Model S makes each musical element and each layer sound better fleshed out and more palpable.

This also ties back to what I mentioned with resolution, detail, and that sense of focus and clarity. The easiest way I found to compare the two DACs was with drums, especially the sort of drums with open bottoms like bongos (unfortunately, I’m not a drum expert and couldn’t tell you what sort exactly I was listening to with my music selections). The Theta Gen. Va did a great job presenting the tone, texture, and detail of someone beating on the skin of the drum. I could sense the tautness of the skin and texture of someone moving their hand across it. The Model S, on the other hand, did a better job making the drum sound like it occupied a three dimensional space due to the sense of air, body, and resonance it presented. I could better hear that low-pitched air and reverb come out from the drum, whereas the Theta Gen. Va made it sound a bit less wholesome, as if all there was to the drum was the skin on top. Of course, I’m exaggerating here to get my point across and thought both DACs did well. Now, personally, the Model S made it seem more real and palpable to me, but I know some would instead be more drawn in with how the Theta represented the drums. I just have to go with what moves me best!

Granted, this exact sense of body I’m referring to on the Model S can ever so slightly get in the way during the most complex passages. For the most part, the Model S cuts through the crap, never gets congested or starts tripping over itself, and retains its composure throughout in terms of detail, texture, separation, and so on, but it has rare moments where it looks like it might start to sweat just a little bit. So, there are moments that, while I prefer the palpability of the Model S over the Theta Gen. Va, the Theta simply offers better layering and separation and, as such, keeps some of the texture intact for each musical element when the going gets tough. It’s not that the Model S seems to really struggle or drop the information, you just have to listen a wee bit closer to extract it. Like I said, though, this is rather rare, and not a huge difference at that.

On the other hand, and this has almost always been the case to my ears with non-oversampling DACs, the Model S does a bit better with micro-dynamics and micro-blackness. Those quick on/off shifts in between the notes. When it needs to, the Model S stops on a dime, whereas the Theta Gen. Va retains that very slight bit of transient smear I hear from most oversampling DACs. Now, could I be completely crazy and imagining things because that’s what I want to hear? Possibly. I can’t rule it out, but I also believe that I hear what I hear. And, as I said, this isn’t always the case. The Classé DAC-1 was an exception and could also stop on a dime despite being an oversampled DAC.

From a tone and timbre perspective, I quite enjoy the Model S. It’s a hair warmer than the Theta Gen. Va, and while that works perfectly for me, it might not for others. As mentioned earlier, the Model S does a nice job portraying the music in an honest manner. The Metrums, in comparison, were a bit more agile and “earthy” sounding to my ears, based on memory. With redbook content, and depending on the USB->SPDIF converter, the Model S can be a touch soft and rolled sounding like any other non-oversampling DAC. With a good SPDIF converter and/or high resolution content, it noticeably sharpens up. But, even with a good SPDIF converter and high resolution content, the Model S never sounds etched or hard while still not being plastic and glossed-over like a Metrum. It makes for a natural, fatigue free listen without being boring.

What I like in particular is that the Model S seems to do a good job with details, textures, and timbre from top to bottom instead of having some good and some bad spots. I particularly like its midrange portrayal. Of course, it has its tradeoffs. For example, I love how it portrays someone finger picking a nylon guitar. It sounds just like it should to me based on hearing people play and playing nylon guitars myself. Most oversampling DACs portray this as a bit too sharp and thin, almost as if the guitarist is using a pick instead of their fingers or as if the strings were a mixture of nylon and metal. Most other non-oversampling DACs will make it sound too dull and too much like plastic. The Model S hits that sweet spot for me. I also absolutely love vocals on the Model S, even more so than I did from the Metrums. They sound incredibly real to me. Again, palpable and fleshed out. Now, like basically every other non-oversampling DAC, they do make things like cymbals sound more polite than they do in real life. That’s what I mean by tradeoffs. In a sense, I just go with what bothers me the least, and it just so happens that having a slightly more polite nature up top does help with long listening sessions and my strange ability to get fatigued from nearly anything. (Again, high resolution content fixes the top end right up and is still more pleasing to me than anything oversampled.)

Lastly, the Model S is quite dynamic and lively sounding. I had concerns before listening, because so many people online with varying TDA1541A experience said it’s a polite sounding chip or focused on the midrange. Very rarely did you hear anyone commenting on it being incredibly dynamic, lively, or having much bass slam. Well, whatever Pedja has done with the Model S, it has certainly paid off. I still have yet to hear a DAC that is quite as dynamic and slamming as the Classé DAC-1 (and perhaps it was colored in that regard), but I can at least say the Model S and Theta Gen. Va more or less seem to be on-par in terms of dynamics. It’s very engaging to listen to and always keeps its composure.

Final Notes

All things said and done, the Model S basically checked all of the boxes I wanted in a non-oversampling DAC. Even after all my research and various DAC experiments, it was still a big risk for me to spend what money I did on a relatively unknown DAC from a different country. Sure, it looked good, and I figured I would like it, but what if it didn’t live up to all of my expectations? Thankfully, I absolutely love listening to this DAC. It’s not even a contest compared to the Metrum Hex or similar. Considering I paid about the same for the base Hex model and the fully upgraded Model S, I cannot at all recommend the Metrum Hex at its price point. The Model S, at least, continues to show what it’s capable of the more I try different and better gear with it.

Now, perhaps something like the Metrum Pavane or a TotalDAC would be a different story. Both are just too expensive for me, and I must say that the Pavane’s THD and other technical/hardware specifications concern me if it is indeed supposed to be so accurate and linear for both 16 and 24 bits. I’m sure it’s better than the Hex, and probably noticeably so, but at somewhere around $5K, I’m cautious. Then you have the TotalDAC, which may not have better than 14-bit accuracy. They also enter the ludicrous price range. I’ll stick with my Model S, thank you very much.

I also have no doubts that the Schiit Yggdrasil is a more technically proficient DAC than the Model S. And given my experience with DACs that Mike Moffat has touched, I would probably enjoy the Yggdrasil. This just all goes back to my preference for a non-oversampled sound. Both oversampled and non-oversampled designs have tradeoffs to my ears. Non-oversampled just resonates with me more, assuming strong technical performance and a good design. So, the Model S fits what I wanted perfectly, and if you like a non-oversampled sound, you really owe it to yourself to try one out. If you prefer an oversampled sound, then, hey, get a Yggdrasil or keep an eye out for trickle down tech to Schiit’s cheaper DACs.

I'll post my basic and modest Model S measurements in the near future. I have the data, it’s just not compiled or sorted. I will give a bit of a teaser by saying it does quite well in almost all regards.

Too Long; Didn’t Read

- Probably the best non-oversampled DAC I’ve heard by far. Metrum Hex is no competition. Stuff like Pavane and TotalDAC are too expensive, so I can’t compare. I’m also still not 100% sold on the Pavane’s claimed accuracy/linearity and am worried about claims TotalDAC may be 14-bit ENOB. TDA1541A is rated at 16-bit, typ. 0.5LSB/max 1LSB, has great specs otherwise, and Model S seems to almost maximize its performance even though it’s a NOS DAC.

- Very good tone. Slightly warmer than Theta Gen. Va but still does a good job honestly portraying unique recording and mastering characteristics of various albums.

- Much larger soundstage and more air than I’d expect from a NOS DAC. Theta does a bit better here, and has better layering too, but the musical elements and layering on Model S sound more fleshed out and 3D to me. More palpable.

- Model S doesn’t gloss over details like a Metrum. Actually very resolving for NOS and doesn’t leave me wanting when I switch back to it after listening to the Theta since it gets rather close to that. Good texture and very rarely starts to get any sense of congestion or strain. There is some focus and clarity to gain by going with something like the Theta, but this gap can be bridged on the Model S with a really nice SPDIF converter and/or high resolution content. High resolution content in particular fixes that slight rolled sense you get from redbook on NOS and puts the DACs on basically even footing with some differing traits.

- Very dynamic and engaging to listen to. Plenty of bass heft and slam. Never dull, murky, or grey sounding. Never hard or artificial sounding. Fatigue free and natural sounding not because it takes away but because it doesn’t add crap.

- Relatively fast. Stops on a dime with micro-dynamics and micro-blackness. Great sense of rhythm.

- If you were on the fence with something like the Metrum (i.e. not resolving, too agile), you might find a lot to like with the Model S. If you really hate the NOS sound, this probably won’t change your mind. I’m not going to delude myself into thinking most aren’t better served by something like the Yggdrasil, but if you want a truly excellent, TOTL NOS DAC, you really ought to consider this Model S. It has been perfect for me.

(Credit goes to Audial and Pedja for some of this info and the picture attached to the review.)

Measurements

See here: http://www.changstar.com/index.php/topic,2525.0.html

[Priidik]

Thanks. I read it all 

Which phones, amp(s)?

[Hands]

Stock and modded HD650, modded PM-2, little bit of HD800, little bit of Slants. Valhalla 2 with stock and nice Telefunken tubes, Singlepower MPX3 with some Sylvania chrome tops, Leck. UHA-6S MKII with Sparkos discrete opamps.

[thegunner100]

GJ Hans! I kinda glossed over the sound impressions since you've already told me about it before. Maybe a Yggy vs Model S comparison will be available in the future! 

[slaine]

How do you get pricing? Is it email only? I can't seem to find the pricing anyway.

[kothganesh]

How do you get pricing? Is it email only? I can't seem to find the pricing anyway.

its on their website... 2200 for the USB DAC and 1900 for non-USB...euros not dollars

[Hands]

The price list is on the "sales" portion of the site. Prices are the same for the MkII SPDIF and USB models. 1900 € without output transformers and 2200 € with output transformers. I went for the latter option, though it looks like prices have gone up slightly since then. I paid 2100 € for mine, though, technically, I paid more in USD at the time due to the conversion rate than what it would cost now in USD.

[madaboutaudio]

Excellent review and detailed analysis of the sound. On the sound of NOS, you are so spot on. Wonder why did the dac engineers decided that oversampling & upsampling was the "solution" when it made some critical aspects of the music sound worse off, especially on the imaging side of things...

I wonder how does this compare to Schitt Yggdrasil.

[Hands]

Measurements now available here: http://www.changstar.com/index.php/topic,2525.0.html