Saturday, 16 October 2021

REVIEW: Topping D90SE DAC - Part III: Subjective Impressions, AMPT, and Summary. A few words about "warmth" and "accuracy" in sound systems.

This is the final part of my trilogy review/evaluation of the Topping D90SE DAC (see Part I, and Part II for the objective testing results).

As we end off, let's close with a discussion of subjective impressions developed over many evenings of listening (about a month or so). I've described the soundroom previously in some detail and for the two channel system, the main amplifier is my home-assembled Hypex nCore NC252MP, main front speakers are the Paradigm Reference Signature S8 v3, and dual subs (main Paradigm SUB1 with the little Polk PSW111). The Topping D90SE feeds my Emotiva XSP-1 preamp. All cabling is balanced XLR from DAC to amplifier (no need to fret, inexpensive Monoprice Stage Right XLR runs are all ya need, feel free to spend more on cables if you see fit of course).

I listened both with and without DSP room correction activated; generally I prefer with DSP. Knowing the frequency response characteristics (flat to 20kHz) and setting to "Fast Roll-off Linear" (Filter 5) allows me to use the same DSP settings for my room interchangeably with the RME ADI-2 Pro FS Black Edition without need to remeasure (equivalent frequency and time-domain performance). As discussed before, one of the benefits of a high quality DAC is that we can have even more headrooom with DSP processing whether it be full room correction or applying volume normalization like ReplayGain.

For a size comparison, here's a peek at a stack of DACs in the sound room with the Raspberry Pi 4 "Touch" streamer on top:

DAC Stack: TEAC UD-501 (2013), Topping D90SE (2021), RME ADI-2 Pro FS R Black Edition (2020). Notice the Topping is about the same width as the RME, but deeper and flatter.

A. A few (obvious) comments about subjective impressions...

Before I begin talking about how the D90SE sounds to me, I think it's important to just say a few things as hopefully obvious reminders by now. Subjective listening and write-ups can be a bit tricky not only because these are not blinded (so they're not controlled) but the comments said can be construed as being "authoritative" yet in reality are just ephemeral impressions and generally cannot be taken as "fact". This is why I always start reviews with the objective tests - that's more apt to be factual, reflect the true performance, than flights of fancy and expressed emotions.

IMO, unless a DAC these days has serious deficits that mess up sound in an obvious fashion (which also means it can be measured and verified), I think it's important to just be mindful of human limitations to perception and not think that subjective opinions need be held with conviction. Listening impressions are generally "testimonies", nothing more. While they might be accurate, always be mindful if there are undertones to the message and "read between the lines" for potential bias. Be mindful of whether the device being testified about came as a result of a relationship with the manufacturer. Is this a (potentially longterm) "loan"? Is the review produced with the intent of monetization? The usual caveat emptor rules apply as with any kind of potential advertisement in an unfiltered online world.

Many times, I remember walking away after a listening session with a negative impression of a certain device and a couple of nights later experience the opposite using the same hardware. Did the hardware "break in"? (A convenient explanation with many audiophiles over the years believing that hundreds of hours of break-in is needed for solid state devices or even passive components like cables!) No, I bet more than 9 times out of 10, it's because I've changed. Perhaps better mood, less stress, better night's sleep, less sinus congestion, ear wax shifted; all manners of potential changes in the cognitive, emotional, and biological machinery affecting the human mind.

I seriously hope that readers, viewers, reviewers all can acknowledge these simple facts about being human so as to put subjective claims and reviews in the right context. As far as I am aware, there are no "professional" listeners doing audiophile subjective review write-ups where their main career is as a meticulously trained listener. This idea is even more important when we're talking about things with little potential mechanistic effects - cables, bit-perfect digital streamers, bit-perfect software, etc...

While I do have a playlist of songs I can use as "standards" when listening to new components, it would be rather boring I think to just spend time discussing a handful of tracks or albums; the same ones each time I stick a component in the system!

Instead, it's much more fun to run through a number of albums I've bought recently or revisited and make some comments about the sound of the system and discuss the albums which some readers might also be interested in. That's what I'll do here with an eye to considering the D90SE's performance.

B. So how does the Topping D90SE sound? Let's have a listen to it in the sound system...

As usual, I listen to a device before I run measurements for a couple of evenings before putting it on the test bench. This allows me to have a listen to it brand new without any "break in". Since I know some audiophiles have a strong belief in "break-in", I purposely let the DAC play on loop with a playlist of rock/pop tracks for about 70 hours after the objective testing before doing more critical listening over the next few weeks. The discussions below were primarily with albums I listened to after that 70-hour "break-in" (no, I do not believe this made any difference to the sound).

First up, I received a couple of new live albums the other week; Donald Fagen's The Nightfly: Live (2021, DR10) with tracks recorded at the Orpheum Theater Boston and the Beacon Theater NYC during shows back in 2019 (pre-pandemic), and Tears For Fears Live At Massey Hall, Toronto 1985 (2021, DR8).

The sound quality from live recordings can be hit and miss. It's nice on recordings like this to listen for the atmosphere in the live venue. Can we hear that sense of "space" and how the crowd noises come across to extend that feeling that one is "present" at the time of the recording?

For the Donald Fagen album, I was certainly impressed by how "clean" the live recording sounded! Right from the start, in my soundroom the crowd noises in "I.G.Y." surrounds me and with the D90SE playback. I can hear the low level background noise of the venue and the audience, even picking out individuals in the crowd. Like with Becker/Fagen Steely Dan albums in general, the studio recording of The Nightfly was immaculately produced. It's quite impressive how good the live band sounded here compared to the studio work in the original album! Notice that the sound of the live recording is not as "sizzly" as the early digital recording. The DAC is able to convey the nice deep kick drum on "The Goodbye Look" and wonderful pacing on "Green Flower Street" (percussion, xylophone). Donald Fagen would have been around 71 years old when the live recordings happened and his voice is still quite strong on these recordings. Overall, an enjoyable album if you like The Nightfly.

The Tears For Fears live album is a very different beast. This is an analogue tape recording done back in 1985 and then remastered for release, in the process it has obviously been moderately compressed to a DR8 value to sound relatively "modern". 

Notice the peak-limiting applied to this live recording with little stand-out dynamics. This implies a loss in some of the nuances of the performance and this is what I hear with the playback. It's alright, the spatiality of the crowd cheers are there and there's a sense of 3D space with the band. But there's a bit of harshness in the sound, the tonality is tipped up and not as neutral especially on Roland Orzabal's vocals compared to The Nightfly Live. I've always loved the deliciously dark lyrics wrapped up in a lively, almost joyful package in "Mad Word" which sounds great in this performance complete with '80s sax. "Broken" provided an example of rendering of spatial depth through the sound system where one can imagine facing the stage, a crowd standing/sitting around you, lead vocals a bit closer, percussion in the rear, and electric guitar left front, keyboard stage right. The complex mix of instruments and vocals easily differentiated through the digital playback.

In the top picture showing the "DAC stack", notice the black cover image of the album I was playing at the time - Kanye West's Donda (2021, DR6). Admittedly I'm not a big fan of rap or Kanye. This album named after his deceased mother unfortunately is a bit of a mess IMO. Massive amount of religious allusions in the music but it's simply not "deep" or particularly insightful. Rather, we're looking at superficial allusions and pop references thrown in here and there.

Yeah, the usual themes are there including poverty, crime, drugs, suicide, disrespect, vague thoughts around redemption through faith, etc... I certainly have no problems with these themes, but it's just not done well and I don't think West's antics over the years have helped create a sense that these ideas are at all genuinely expressed. For example, I kind of enjoyed "Jesus Lord" with its multiple layers and themes sonically, but it's also filled with clichés and pop references which will not stand the test of time. Having said this, there are some enjoyable "sounds" to be heard on this album. The intense bass thumps on "God Breathed", digital stereo expansion effects, and shift towards an ethereal layer at 45 seconds with background chorus sounds pretty cool. IMO, there's nothing "masterpiece" about this and comes across as the work of a superficial hedonist with pretensions of emotional and spiritual depths. I doubt I'll be returning to this album much, far from The College Dropout or Late Registration of West's younger years when he made more sense. 

Feel free to have a different take of course, such is the subjectivity of music appreciation. Highly compressed and synthetically produced albums like these certainly sound fine through the D90SE.

Next up, to clear my auditory palate, I had a listen to TakéDaké & John Kaizan Neptune's Asian Roots (1998, DR14). The track "Japanese Roots" is part of my AMPT test recording and I think is just a wonderful "audiophile"-type track to highlight dynamics, decay, soundstage, and tonality of an acoustic performance. I love the articulation and mastery of the Shakuhachi flute shown by Neptune and the precise "attacks" on the percussion instruments are "to die for" as some audiophiles might say. ;-)

It's recordings like these where you have subtle nuances ("Angklung Journey" is a nice example) along with impactful dynamics, and "space" in between notes where one can appreciate the purity of the silence one can achieve in a high-fidelity sound system where a quality DAC like the D90SE excels. This reminds me of Miles Davis' comment that "in music, silence is more importance than sound". This ability to achieve at times explosive dynamics then pure unperturbed silence is like achieving infinite "black level" with the best video displays in a home theater.

As a general rule, a good sound room, low ambient noise level, clean yet powerful amps, good quality speakers with full frequency response and balanced cabling are all highly recommended, if not essential, to extracting the best of the technical abilities of a DAC like the Topping D90SE.

Also within the "world music" theme but more along pop lines, Ladysmith Black Mambazo's The Best of: The Star and the Wiseman (1998, DR12) is a lovely, uplifting album.

The human voice provides a great test of sound system tonality especially with multiple a cappella overlaid parts on the title track "Inkanyezi Nezazi" as a great example. Another standout track for me is "Rain, Rain Beautiful Rain" with its dynamic nuances, sudden spontaneous utterances, and wide soundstage presentation. This sounds fantastic with the D90SE. Highly natural rendering of the voices without harshness or artificiality, enough detail to pick out individuals in the choir.

On one of the evenings, I enjoyed Hélène Grimaud's Memory (2018, DR12). These were piano performances at the Himmelfahrtskirche Sendling church in Munich. No, nothing tremendously innovative or challenging here dynamically for the DAC to reproduce although the analytical mind could focus on the quality of the attack/decay/sustain/release envelope of the piano notes and recognize the technical proficiency of this recording and the D90SE DAC's ability to retrieve these temporal and microdynamic details.

Rather than being analytical, I listened to the album after a busy day at work with the intent to meditatively flow with the melodies and rhythms, taking the opportunity for the spirit to connect with the music and explore whether the sound system components are able to take me to that place of emotional and spiritual engagement. Yup, no problem. These are some of the best times I've had in the sound room and the Topping D90SE integrated into my system had no problems bringing me there.

Let me mention The Shawshank Redemption (1994) which just came out on 4K UHD Blu-ray and was a wonderful movie to share with the family the other night (trivia - Shawshank is the highest rated movie on at 9.3/10 with 2.5M votes).

The soundtrack and score by Thomas Newman (1994, DR11) is likewise a wonderful complement to the movie which I revisited with the Topping D90SE. There's a nice mix of musical styles in this soundtrack ranging from folk Americana like "May" and "Workfield", to the orchestral "Shawshank Prison [Stoic Theme]" and "So Was Red", to slow piano-driven "New Fish", the melancholic "Brooks Was Here", and "Zihuatanejo", to mixed acoustic-synthetic tracks like the quizzical "Rock Hammer". Interspersed among the score are a few vocal tracks like The Ink Spots' "golden era" sentimental "If I Didn't Care" (mono, 1939), honky tonk Hank Williams' "Lovesick Blues" (mono, 1949), and the soaring exotic operatic vocals on Mozart's "The Marriage of Figaro: Duettino-Sull'aria" done by Deutsche Opera Berlin in this album. The sound system negotiated these tracks with grace and accuracy. There's surprisingly deep bass on "His Judgement Cometh" and "Shawshank Redemption" that woke up the subwoofers and I can't help but be moved by the simple, sweet, graceful, majesty of "Suds on the Roof".

A beautifully deep movie, and likewise varied soundtrack.

Finally, I want to thank Mario Martinez of Play Classics for his generosity to the folks on Audiophile Style. As I've expressed over the years (for example back in 2014), high quality audio production is an essential part of audiophile experience which I think we should be knowledgeable about. This cannot be understated as audiophiles because no matter how great we feel our hardware might be, or how awesome the acoustics in our sound room might be, unless we can feed speakers in that room (or our headphones) with seriously good sounding recordings, we might as well be driving a luxury car down country roads, never experiencing the comfort and smoothness of the ride.

I applaud engineers like Mario and his team in the incessant desire to optimize his recording / producing / mastering techniques with the intent to create recordings of utmost "truth" to the artistic performance. While there might not be an actual "performance" when it comes to complex studio productions or electronic music, with acoustic recordings of classical music with actual performers in a real "space", there is at least an opportunity to capture as "true" as possible in 2-channels what the performance might have sounded like if one were present in that space. Mario can say this better in his thread on Audiophile Style where he not only discusses his intent, the various iterations of his "Truthful Recording Technology" method, but also offered readers an opportunity to try out the recordings.

I took up his offer and as you can see in the picture above, downloaded the Olga Yakushina & Elizaveta Yaroshinskaya Around The World with Kreisler for a "spin" (24/96, 2021, DR14, available on Spotify, Apple Music as well).

Let's just start by saying that this is the "real deal" in terms of a "hi-res" recording:

Compare the look of this waveform plot to the Tears For Fears as an example of what "natural" looks like (and also sounds like).

Notice the dynamics on the waveform across the track, natural extensions toward 0dBFS without peak limiting or, God forbid, digital clipping. Note the presence of the violin harmonics up to 32kHz (beyond CD-resolution 22.05kHz) on the FFT, naturally rolling off as recorded by the microphone; beyond which we have a noise floor below -120dBFS and no unusual ultrasonic anomalies. This is the kind of quality I look for if I were to keep a high-res recording in my music archive rather than downsampling as I have often done (discussed being "Post Hi-Res Audio" here).

So how does an album like this sound with the very accurate Topping D90SE as source of the analogue signal? In a word fabulous on my system and soundroom.

On this album are 13 "violin and piano miniatures" (<6 minutes each) channeling the performances by the celebrated violinist Fritz Kreisler (1875-1962) who back in the early 20th Century performed regularly with friend Sergei Rachmaninov as his piano partner. The performances here by Yakushina (violin) & Yaroshinskaya (piano) are expressive, dynamic, and energetic. Emotionally the music is joyful and uplifting; a real pleasure when you just don't have time for full symphonic compositions!

As audiophiles who appreciate the "sound" of the recordings, we can explore the lush tonality, spatial "bloom" of the two complex instruments. Have a listen to the warmth and appropriate stridency** of the violin attacks, the complex dynamic envelope of the piano, surround ourselves within the "soundstage" captured in this recording. For the most part we have the violin placed to the left, piano on the right, and can experience the notable relative depth between the instruments with the piano sounding further back beyond the dimensions of my room and violin closer to the listener "sweet spot". Make sure to have a listen to probably the 3 most well-known of Kreisler's compositions "Liebesfreud" (Love's Joy), "Liebesleid" (Love's Sorrow), and "Schön Rosmarin" (Lovely Rosemary) in series (the Alt-Wiener Tanzweisen "Old Viennese Melodies"). The playful interchange between piano and violin on "Poupée Valsante" (Dancing Doll), speed and technicality of "Tambourin Chinois Op. 3" are also highlights for me. Oh yes, for the pop music listeners, you'll recognize "Preghiera" in Eric Carmen's "All By Myself" (check it out on The Best of Eric Carmen).

Thanks again Mario and I wish you many opportunities to bring your recording skills and experience to other amazing projects ahead! The trick is really to capture the performers and performances that will appeal to audiences at a time when there's just a lot of music out there.

I mentioned in Part II that the ESS DAC's DSD performance had variable noise floor issues depending on DSD rate. I did not bother listening to DSD128+ simply because I don't have any music of the "hi-res" DSD variety unless I did something like upsample PCM (I don't see the point in doing so)! However, I did listen to some DSD64 tracks including selections from SACD rips of Jorma Kaukonen's Blue Country Heart (2002), Telarc's 1812 Overture (2001), and Gaudeamus/Paul Halley's Sacred Feast (2000). I selected these as original DSD recordings that highlight different elements of sound - acoustic vocals, dynamic classical, and complex choral recorded in a resonant 3D space. The D90SE sounds great with these especially the mid-range fullness of the Gaudeamus. Even if not objectively as spectacular in resolution as PCM, I can't complain!

All-in-all, I think the Topping D90SE integrated well into my sound system, one aiming for neutrality and transparency. To me, there is no such thing as sounding "too hi-fi". Bring on the details as far as my hearing allows, and if an album sucks because it was recorded/produced/mastered poorly, I want to know about that too - because in my worldview that's what honest audiophile-quality sound systems do! Not just try to make everything sound "nice" (which is impossible anyway).

C. Topping D90SE AMPT

So, you want to have a listen to the output quality of the Topping D90SE with CD-resolution playback? Well, here you go:

Topping D90SE AMPT (44.1kHz material recorded as 24/96 using RME ADI-2 Pro FS)

While the RME ADI-2 Pro FS ADC is not capable of the precision of the E1DA Cosmos ADC when running measurements, it's easily able to record the output with better resolution than the source "standard resolution" 16/44.1 music fed into the D90SE. Also, for standardization of the recordings, this ADC will be a controlled variable.

Feel free to compare with the AMPT recordings from other DACs!

D. Final Conclusions

So, what we have here in the Topping D90SE DAC (~US$900) is the "best" measuring PCM DAC at this time in history. It has the lowest THD+N/highest SINAD (better than -120dB THD+N) demonstrated by various objective reports over the last number of months; consistent with my findings last week also. As usual, best measuring does not mean subjectively the "most euphonic" for all listeners. Euphonic sound does not need to nor should it necessarily follow technical ideals (take the Pass Labs SIT-2 amplifier for example).

I specifically identified "PCM" above with regards to the D90SE (using ESS ES9038 Pro DAC internally) because I generally find that AKM-based DACs are better technically with DSD playback. Some, like the AKM-based RME ADI-2 devices are capable of "DSD Direct" which bypasses an extra layer of internal sigma-delta processing that DSD "purists" likely will find desirable.

A few months ago, I had already said in my Topping D10s review that I believe many DACs are "perceptibly perfect". Well, since I also said the Topping D10 Balanced was even "more perceptibly perfect", then this beast is clearly "ultra perceptibly perfect" - more than likely exceeding ideal human auditory limits by many times.

To be honest (which is what I try to do here always ;-), I think this is the kind of DAC one buys for the sake of owning "the best", cleanest, most "accurate" reproduction of the source digital signal. Yes, of course one enjoys the music coming out of it and can get lost in the reveries of the most moving of musical compositions. But these days, you don't need to achieve this level of performance or even spend anywhere near US$1000 for an exceptional DAC! No friends, just like expensive cables, a DAC like this brings with it an intellectual "non-utilitarian function" of having something with "bragging rights" as being the objective "best" - at a reasonable price.

Do I need this DAC? No, personally, of course not. I can enjoy music and get lost in it with much less expensive devices. But it's cool. Sometimes we buy something because it's cool. If my wife asks why I have another DAC in the soundroom, I tell her because "it's cool". No need to derive excuses to fellow audiophiles that "the microdynamics are better", or "it makes the piano tonality so much more real"!

One good side effect once we've recognized that a great, or even "humanly perfect" sounding DAC is not difficult to find/buy is that we can exit the merry-go-round of always talking about "sound quality" and can be free of pretentious subjective claims. Let's talk about features like DSP that can be applied, or let's talk about how cool it is to have better "look and feel" qualities like a nice screen or physical controls. Sound quality IMO is better determined by typically "good enough" measurements and we just need to choose among "more than good enough sounding" devices that fits our budget, look great, and have all the features we desire.

That's all I have to say about his DAC.


** In audiophile writings, one often sees a desire for a "warm" sounding system. While this can be enjoyable when appropriate, that's not what I look for in a high-fidelity system. Imagine the horror (at least boredom) of listening to Guns N' Roses' "Welcome To The Jungle" on a "warm" (and comfortably fuzzy?) audiophile system. Likewise, a "warm" and perhaps subjectively "laid back" sound on Laurie Anderson's "The Day The Devil" (on Strange Angels) probably would not sound right.

How about the term "stridency" which doesn't just mean a pejorative "loud, harsh, grating" sound, but also "commanding attention by a loud or obtrusive quality". What good is a "high fidelity" system if it's not able to convey the obtrusive stridency of the devilish tritone at the start of Saint-Saëns' "Danse macabre Op. 40" (check out Rachel Barton Pine's version on Instrument of the Devil)? Music through good sound systems need to be able to express the full range of human emotions including the dysphonic perception of anger, frustration, and horror. "Warm" and "euphonic" sound as a general target is not what I would personally aim for - "neutrality" and "transparency" would be my preferred adjectives.

When audiophiles argue about "what is accuracy" (like this recent Steve Guttenberg video), they typically fail to recognize that the only thing we can say a high-fidelity system can be accurate to is the source material; not what was supposedly heard in the studio control room, nor the live performance, nor even what the artists think they heard (for all we know, they could have used earbuds on an iPhone to listen to the final CD). As technically astute audiophiles, we know exactly how a digital audio file should be reproduced through a high-fidelity system without distortions or added noise - that technical ideal is what we should be aiming for, and can be achieved with high resolution devices in an acoustically controlled space.

I believe that whether a recording is capable of sounding "real" or not was never really something the audiophile could control! All we can do is provide the best environment (room) and collection of electronics and transducers to extract the sound without causing audible "damage" (distortions, adding noise, colorations). Within reason, there's nothing wrong with putting a system together "tuned" to make favourite recordings sound more "real", but if what we're doing is catering to the idiosyncrasies of a handful of recordings, those choices likely will make other recordings sound less pleasant.

As much as we celebrate the abilities of known hardware designers and their speakers or amplifiers, I think audiophiles should champion the audio engineers who truly care about the production side and know how to bring forth natural tonality and nuances in the music encoded whether in PCM or DSD. IMO, these audio engineers and their work are the products which determine more of the sound quality we love than even hardware devices which generally perform very well these days in the 21st Century at lower costs than any time in history!

In the big picture, audiophile Nirvana approaches when we have equipment of such fidelity as to achieve essentially perfect reproduction of the music content we send it, and audio content itself captured and encoded in such a way as to highlight the beauty of the music that the audiophile can find full satisfaction in. I believe the former is easier to achieve than the latter, yet we don't seem to discuss the latter enough.

[If we think about it, of course within the hardware audiophile forums, magazines, and YouTube channels, we have a tendency to discuss equipment more than audio engineering or the mastering of music itself. We have self-selected to be interested in the products as a consumer group. And the purpose of advertising dollars is to drive up that desire for hardware acquisition.]


Well guys and gals, it has been a bit of a whirlwind these last few weeks finishing off the Topping D90SE DAC review and measurements as well as laying down the framework of the SoX-DSD testing, while getting acquainted with the E1DA Cosmos ADC! Time to take a little break and "digest" the material and also just enjoy the music more. ;-)

I need to attend to a few work and extracurricular activities over the next little bit...

Stay safe, and I hope you're all enjoying great sound.


  1. A series of articles Thank you for your hard work.
    I read it very interestingly. thank you.

    1. Thanks oagstes,
      A pleasure...

    2. Hello.Thanks for recording the demo sound!
      Topping D90SE still lost on the naturalness of the old RME ADI-2 Pro FS R.
      RME ADI-2 is better for my subjective taste!
      1.from track time 7.40 there is no bass depth.The stage is flat.
      2. from track 5.05, the vocals sound thin and not natural.Synthetic voice.

      What makes the D90SE better is the better audibility of quiet background sounds.

    3. Thanks for the feedback Geprg,
      Great that you had a chance to download and listen. We can certainly say that technically we can measure and see these differences. I'm happy that we can each listen for ourselves!

  2. Thanks Archimago,

    It is great that you got the Cosmos ADC in time and did detailed DSD vs PCM analyses in time before making a conclusion about the D90SE. Also, excellent and easy to understand "infographic" grade illustrations in 44.1k PCM vs DSD upsampling with the RME's DAC output, clearly showed the DSD noise vs imaging artifacts in 44.1k PCM, given the fact that AKM's "sharp" PCM filters only have 100dB attenuation.

    For DSD, I believe even your UD-501 can achieve better performance than any ES9038 based DAC. Of course, if you are still curious, CS43198 (e.g. Topping D30Pro) seems to have DSD Direct support as well.

    In fact, I have no strong opinion about music released in DSD, but I do think that DSD upsampling in the playback chain is a waste of time. People have to upsample to at least DSD256 to make the ultrasonic noise floor comparable to PCM due the slowness of typical on-chip DSD filters, it is far more resource intensive than PCM resampling to for example, DXD rate, which can also greatly reduce imaging, plus there is zero rise in ultrasonic noise floor at least up to several hundred kHz.

    Also good to know that MC_RME is also evaluating the Cosmos ADC, which is a good sign for the advancement of future studio interfaces. Of course, I still hate ESS's silly marketing about "native" DSD1024 support, and MQA integration in some of their products.

    1. Hi Bennet,
      Thanks for the feedback man. Yeah, at some point I might just have a look at the TEAC UD-501's DSD again (TI PCM1795 chips) and how it compares with the SoX-DSD test signals.

      Agree, I know some guys really like their DSD upsampling with HQPlayer and other software, often with high-powered i9 CPUs and GPUs. I of course tried some of this back in 2019:

      Yeah, that can certainly achieve some excellent quality filtering but I would not put too much stock in achieving significant audible differences given the amount of investment in processing power!

      Yeah, would be interesting to hear of MC_RME's take on the ES9822 and the evolution of the ADI-2 Pro series now that the E1DA Cosmos ADC can demonstrate this level of ADC performance!

      Well, I guess with the D90SE there's that DSD1024 I2S input option but given what I'm seeing up to DSD512, there's really likely no point. Interesting to hear that the CS43198 does DSD direct - I see up to DSD256 - which is certainly good enough. Will keep an eye open for this DAC ;-).

    2. BTW Archimago,

      It seems that D90SE's AMPT is not using your favorite linear phase fast filter? Given the 22kHz stopband.

      Also, JohnYang1997's UD-501 measurement seems to have significantly better THD+N than yours, can you get a better result if you slightly reduce the digital volume?

      Regarding CS43198, if you look at the DSD filter plot, it has 100dB attenuation at 180kHz and over 180dB beyond 500kHz (on the "Patented DSD Processor" instead of Direct DSD?) ESS's 50kHz IIR filter on the other hand is only 60dB at 500kHz. PCM1795's DSD filters are pretty weak as well.

    3. Hey Bennet,
      Oh I will have to check the AMPT out and maybe do a fresh one opportunity permitting. I did it during some early testing so might have used one of the other variants accidentally... Anyhow, I trust the differences will not be too much. ;-)

      Interesting, I see JohnYang1997 got ~113-114 SINAD on the UD-501, XLR out at ~4Vrms. Let me run my unit tonight using the Cosmos ADC's mono mode and see what I get! Don't think I've done a more "proper" reading of the UD-501 with the current equipment yet.

      Cool on the CS43198 DSD filtering. Looks like that would be the best DAC to use if there are any concerns around ultrasonics causing instability in certain amps.

    4. Didn't get a chance to run the UD-501 through testing last night. Will maybe aim for a quick post next weekend with a few more of the tests I use these days and have a second look at a 2013 "classic" DAC. :-)

      I can say that it's certainly better than the T+A DAC 8 ASR measured in that thread, but not sure I'm seeing the level of SINAD reported either! This will also allow me to have a quick peek at the noise level with DSD playback on the TI PCM1795.

    5. Just an update from an engineer friend regarding the CS DACs.

      His warning was that we cannot use just the datasheet information absent of showing the noise floor.

      Also, he pointed out to the Topping D30 Pro measurements at ASR:

      It looks like there's more noise shaping in the PCM playback. My engineer friend was concerned that the sum of the noise could be problematic in pro studio environments. ESS and AKM DACs are better in this regard.

    6. Thanks Archimago,

      In fact I read the same D30Pro review before as well and left comments, for example this one:
      Scarlett 2i2 uses CS4272 codec (DAC+ADC), CS4392 (DAC only) on the EMU0404 USB has similar characteristics. These chips seemed to use multibit modulators up to 128fs.

      However, the noise hump of CS4272 only goes up once and dies out completely.

      Unlike BB's DSD filters:

      Of course, the CS43198 has much lower noise shaping that CS4272. What I can get is that CS43198 is not running at a very high modulator speed (256fs?), but with much more sensible post-filtering, and the IMD vs amplitude plot is also 10dB cleaner than the DX3 Pro (unbalanced)

      So basically what I am seeing is that CS's approach is to use the minimum required resources/power consumption to make the best possible outcome (e.g. 120dB sharp PCM filters), and many good measuring dongle DACs are also using lower-end CS chips. Maybe that's why Topping needed to use 4 CS43198 chips to achieve this level of performance, but on the other hand, the retail price is also significantly lower than their AK4497 DACs.

    7. BTW, here is the bus-powered and unbalanced Sound BlasterX G6 with a single CS43131:

  3. Thanks for this series. Each was fun. You wrote above, "One good side effect, ..., is that we can exit the merry-go-round of always talking about "sound quality" and can be free of pretentious subjective claims." For me there's a slightly different twist on this thought. We can have higher immunity to FOMO emotions when we read about the latest offerings. The D90 should give one a lot of confidence when listening to music. For me that confidence helps to relax and enjoy. Same goes for other gear you've measured. Thanks for the measurements and the human incites.

    1. Hi Doug,
      Thanks for the feedback.

      Yeah, that's certainly the kind of sentiment I wanted to go for. The idea that the audiophile is a neurotic, anxious animal driven by FOMO is probably true in many instances.

      We've been trained by decades of writing, audio shows, and dealers to believe that there's something profoundly mystical about the experience. That science somehow has not been able to measure stuff that's supposed to be there - like our perennial debates about whether or why stuff like power cables make a difference to some!

      I agree that doing a more detailed review and measurements like this can demystify the product and hopefully conveys to the readers that we are still looking at engineered performance of devices of human design and ingenuity. And even in the "best" designs, there may be areas of relative weakness (like DSD here with the DAC chip).

      Indeed, as humans, we all have limitations and with the various Topping DACs, RME, TEAC, Oppo, I can at least try my best to convey that even subjectively, I'm not hearing major differences when played side-by-side. ;-)

      That's what the science predicts and that's what I'm experiencing as well through my speakers and headphones!

  4. 3 very interesting and detailed articles. You have described the D90se as the most faithful, I would like to ask you for your opinion in comparison with the D70s, thanks

    1. Hi Nettyno,
      Based on what I have seen, I think the D70s (dual AK4497) should be excellent. In fact, I would anticipate that it would function very well on the DSD playback side.

      I heard there may be issues with the Topping D90 and use of MQA filter parameters even when not decoding MQA. Hopefully this is not a problem on the D70s which also implements MQA.

  5. Archimago wrote: "I believe that whether a recording is capable of sounding "real" or not was never really something the audiophile could control!"

    I believe I know what you mean, but I think I have to disagree with this.

    Obviously much may depend on the recording. And strictly speaking, if we wanted scientific rigor, the more “real” sounding system would be determined through blind testing, perhaps against real sounds, to see which score closer to the real thing.

    But in the real world, we are dealing with compromises in music reproduction with respect to trying to reproduce “realism.” But even if true realism isn’t attainable, a systems can nonetheless take steps towards or away from sounding “more real.”

    I’ve spent a lot of time trying to get my system to at least mimic aspects of “real voices and instruments” - paying attention to the nature of real sounds, and seeing what aspects I personally seem to care about (since perfect reproduction is impossible) and trying to nudge my system in that direction. I have all sorts of tools to do this, from using my tube amps which add something I find ‘natural’ to the sound, to playing with the settings with my vinyl set up (e.g. impedance settings and VTF can alter the precision, lushness, highs, airiness etc of vinyl playback). And I can play with room acoustics which I do all the time. I can use all these “tools” to slightly, but importantly, increase the sense of “realism” to my ear. When I get things really dialed in drums, sax, voices ect have a more “real sounding” “correct” timbre to my memory of real sound (and all I have to fool is my own memory). Virtually EVERYTHING sounds better when this is the case.

    If things aren’t dialed in...NOTHING really sounds “right” or real to me.

    I have a friend who reviews audio gear so I always listen to whatever mega speakers he has at the time. Some can sound astonishing clear and present...yet nothing at all sounds “right” or “natural” to my ears, I can’t believe any of it. I go home to my system and it just shows me how much I’ve been rewarded in all my efforts because my brain just goes “Yup, this is correct, that’s how a sax sounds” or whatever.

    Again...the only ear I have to fool or “tune” to is my own. Whether someone else would like the decisions I’ve made is another thing. (Though I can say that over the years guest after guest has been astonished at what they describe as the realism of the sound in my room).


    1. Thanks for the note Vaal,
      When I wrote that line I had a feeling there would be some comments directed at it. ;-)

      I certainly get what you're saying. Indeed, what the audiophile CAN do is make sure we optimize the room, set up our speakers to the best of our ability (or at least the limits of the room), get the best combination of electronics we can so that there is the hope of achieving the illusion of a "real" performance, at least evoking the impression of being in the venue when we close our eyes and play a well-recorded acoustic album. That's a lot of stuff we can still do and fun we can still have.

      I used the words "whether a recording is capable of sounding real" because I think before any audiophile even gets a chance to play recordings in their optimized systems, we are *locked* into whatever data is on the disc, or in the file. And unless the production engineering side gets it done right, we're never going to "make a silk purse out of a sow's ear" as the saying goes.

      Unless the musical nuances are retained in the recording (good frequency, dynamics, and time domain characteristics), the necessary hints of the acoustic space embedded, the microphones positioned to capture what's needed for convincing soundstage reconstruction in playback, no matter what we do, the "illusion" would never be achievable.

      Of course, many recordings these days are not meant to sound "real", but rather our sound systems are actually "performing" the audio which was constructed virtually in Pro Tools. For classical and jazz acoustic recordings, I definitely would be listening for that illusion of "real".

  6. I loved your trilogy! Fantastic review(s)... Although I'm not a big fan of referencing sound impressions to various albums,(I know most reviewers do it), I found your measurements and general impressions of the D90SE very informative. Thank you!

    1. A pleasure Rlwings,
      Yeah, I certainly appreciate the comment about the way that reviews are done with reference to albums.

      I'm not sure on the subjective side what options we have apart from bringing up some actual music and expressing that "holistic" impression of sound quality in the context of subjective emotions.

      If folks have an example of a subjective review out there that could work or is more systematic, I'd love to take a look at what could make subjective reviewing better!

  7. Thank you for the review. I honestly like your blogspot:-)

  8. "many recordings these days are not meant to sound "real"". I strongly agree!
    After years of equipment upgrading, you get to the point where your audio system is so much accurate that you expect nothing else but the best when listening to it, but... Which such a high level of performance, you soon discover that quite a large number of recordings are very badly mixed: so deceiving..."all dressed up, but nowhere to go"! Fortunately, when you get a recording of your preferred music that is also superbly recorded, alleluia, this is THE reward that prevents you feeling bad having spent that amount of money on it !

    1. Nice comment DColby,
      Yeah, I for one thinks it's high time we seriously look at the recordings and not make assumptions that there's a "need" to coerce our equipment to be "natural" or "real" sounding with everything.

      Indeed "all dressed up, but nowhere to go" is sadly true with a lot of what I listen to at least. ;-)

  9. Excellent work again, Arch.
    This series could be a model of how equipment reviews _should_ be performed and written. Comprehensive, open-minded, and sensible.
    As for the more philosophical comments, I think that now that we have more or less perfect (as far as human perception goes... not far?) electronics, the focus will naturally move to recordings (as in the comments here), transducers (the weak links in the chain), acoustics, and finally user preference in terms of music recreation.
    I was interested to read in ASR how Amir, the extreme objectivist, finds a particular open reel tape played back on a ~40dB SINAD deck to be a satisfying (reference?) experience. The source is primary.
    Your comments about whether artificial sweeteners enhance or harm some recordings (like the Laurie Anderson track) are interesting... I've certainly had trouble implementing a one-size-fits-all solution (to mix some metaphors.) I hope one day we can just dial in or select a playback configuration that will reflect something like what the engineers in the booth heard, and DSP is taking us closer to that. Of course, we might not like what we hear :)
    Keep it up, sir, I'm sure your posts are weekend highlights for your growing audience.

  10. Regarding the footnote(**) on warm sounding systems, it may not be what you are looking for, but I now understand why other people seek it.

    My components during my formative years consisted of warm-sounding speakers paired with a class AB amp. It made for a great "all-rounder" system on which I enjoyed many genres like jazz, classical, rock, electronica, etc. for two decades. Recently I revamped my system and "upgraded" to neutral speakers and a class D amp, and I must say I miss the warmth of the old system. I certainly have more accuracy and detail (I can hear the HVAC system of a Daneil Barenboim piano concert!), but I'm not sure I really wanted to be distracted by that when listening.

    I have a newfound respect for people whose stated goal is a warm-sounding system. I think it's OK to want to wrap yourself in a warm blankie and enjoy jazz the way you like it. I see the hobby now as striving to create an individualized system by matching components to achieve a desired outcome, much like a gearhead working on a custom car. I measure success by how much I do not want to tweak things any more, and just enjoy the music.