Review: High-end Yamaha receiver versus UltraRendu with Audirvana upsampling

Studio
1

Review: High-end Yamaha receiver versus a dedicated UltraRendu with Audirvana upsampling

Introduction

In my recent review of Audirvana Studio, I documented my journey optimizing upsampling settings to achieve what I described as “magical” sound quality through my UltraRendu → IsoRegen → Benchmark DAC3 chain. Shortly after publishing, an interesting comment appeared from a forum member suggesting I might achieve even better results using my Yamaha RX-A8A receiver’s ES9026PRO DACs:

“Why don’t you use the RX-A8A for high-resolution playback? It uses the ES9026PRO Hyperstream II chipset (2x) supporting up to 32/384kHz and DSD256… you say it sounds awesome [for surround SACDs]… You would gain the room correction DSP in concert with up-sampling…You can connect your mono-blocks to the pre-amp outputs…”

The suggestion was intriguing. My Yamaha RX-A8A is indeed a high-quality receiver with impressive specifications, including dual ES9026PRO DAC chips capable of DSD256 and 32-bit/384kHz PCM processing – capabilities that exceed my Benchmark DAC3’s DSD64 limit. Could my carefully optimized multi-box solution be outperformed by a modern integrated receiver?

I love a challenge related to my favorite hobby (listening to music), especially a challenge that might save me and others money or simplify our systems. So I embarked on a methodical listening comparison between these two very different digital playback chains, using identical content and source components. The results were illuminating, sometimes surprising, and provide valuable insights for anyone considering different digital playback configurations with Audirvana.

Test Setup

Common Elements:

  • Mac mini (late 2012) with Audirvana Studio
  • Identical upsampling settings for each comparison
  • Same reference tracks across all tests
  • Similar power cables (high end power cord on the receiver, similar cables and linear power supplies on the IsoRegen and UltraRendu)
  • Same amplifiers
  • Same speakers

Setup A (Reference):

  • Audirvana → UltraRendu (via UPnP/DLNA over EtherRegen switch) → IsoRegen (USB reclocker) → Benchmark DAC3 → Yamaha RX-A8A balanced input
  • SoX upsampler with optimized settings from previous review
  • DSD64 conversion with Type A (4th order) modulator
  • Digital components powered by linear power supplies (LPS)

Setup B (Alternative):

  • Audirvana → Yamaha RX-A8A (via UPnP/DLNA over EtherRegen switch)
  • Various upsampling configurations tested (detailed below)
  • PCM 384kHz/32-bit and DSD64/128/256 tested
  • Both SoX and r8brain upsamplers evaluated

Note that I connected the Benchmark DAC3 of Setup A to the Yamaha receiver’s input, instead of to my mono block amps. This way, either setup users the Yamaha’s pre-amp. So I really tested DAC performance and not the pre-amp or amps.

Key Findings

Sound Character Differences

Benchmark DAC3 (Reference Chain Setup A):

  • Deeper, more three-dimensional soundstage
  • More sustained and impactful bass reproduction
  • “Blacker” background with better contrast between instruments
  • More natural decay of instruments and reverb tails
  • More emotionally engaging overall presentation
  • Superior reproduction of hall ambience and spatial cues

Yamaha RX-A8A (Direct UPnP/DLNA Setup B):

  • Clean, detailed sound with precise imaging
  • Flatter soundstage compared to reference system
  • Less sustained bass with quicker decay on low frequencies
  • Excellent clarity but sometimes with a more analytical character
  • Slightly drier presentation with less natural reverb
  • Very good but less emotionally engaging reproduction

Format Preferences

One of the most surprising findings was how differently each DAC responded to various digital formats:

Benchmark DAC3:

  • Performed best with DSD64 conversion
  • Exhibited that “magical” quality with optimized SoX settings
  • Higher DSD filter orders (5th-8th) reduced naturalness

Yamaha RX-A8A:

  • Performed best with PCM 384kHz/32-bit
  • Noticeably preferred PCM over DSD (at any bit rate)
  • Produced audible relay click when switching between PCM and DSD (indicating different signal paths)
  • r8brain upsampler provided better bass than SoX for this DAC, but overall I still preferred SoX
  • When using DSD, the Filter Type A with 5th Order sounded better than my reference system’s 4th Order. But 6th-8th introduced a very slight “veil” over the music.

Detailed Listening Observations

I used seven reference tracks for consistent evaluation. Here I’ll highlight the most revealing findings from three representative tracks:

“Sleepers Beat Theme” - Ben-Lukas Boysen

This minimalist track with ultra-deep bass and piano revealed stark differences:

Benchmark/DSD64:

  • Deep bass notes sustained naturally with long decay
  • Physical sensation of bass pressure during sustained notes
  • Piano attacks with both precision and natural harmonic character
  • Harp positioned distinctly far behind the piano in the soundstage with natural sparkle
  • Exceptional contrast between notes and silent background

Yamaha/PCM384 (best configuration):

  • Deep bass present but with shorter sustain
  • Less physical impact from bass notes
  • Good piano definition but slightly less dynamic
  • Harp positioned more forward in the soundstage
  • Good but less dimensional presentation overall

Yamaha/DSD256:

  • Bass less sustained than PCM384 version
  • Slightly “lighter” overall tonal balance
  • Less natural piano timbre
  • Flatter soundstage presentation

“Powaqqatsi Anthem Pt. 3” - Philip Glass

This complex orchestral piece with subtle spatial cues was particularly revealing:

Benchmark/DSD64:

  • Snare drum positioned with natural reverb extending deep into the wider soundstage
  • Left-to-right panning synthesizer moves with precise tracking
  • Pan flutes with natural “velvety” texture and air
  • Timpani remains powerful and defined during crescendos
  • Individual instruments maintain separation during complex passages

Yamaha/PCM384 (best configuration):

  • Snare drum positioned a bit more forward with less reverb extension
  • Panning synthesizer movement less distinct
  • Pan flutes clean but with less spatial information
  • Timpani present but less impactful during crescendos
  • Some minor smearing during the most complex passages

Yamaha/DSD128 (with forum-suggested r8brain settings):

  • Slightly veiled presentation overall
  • Less distinct panning synthesizer
  • Slight crackling artifacts in pan flutes before crescendo
  • Timpani lacking full power during climax
  • Generally less transparent than PCM384 version

“Evacuation” from I Am Legend - James Newton Howard

This emotional orchestral/choral piece tests separation and layering capabilities:

Benchmark/DSD64:

  • Opening cellos with natural body and warmth
  • Strings and choir distinctly layered in separate spatial planes
  • Deep bass synthetic effects with power and control
  • “Angelic” female choir with presence, natural roundness and air
  • Timpani positioned naturally at the rear of the soundstage with power during crescendos

Yamaha/PCM384 (best configuration):

  • Good cello definition but less warmth
  • Less distinct separation between choir and orchestra
  • Deep bass present but less physically impactful
  • Female voices slightly drier and more forward
  • Some smearing in central orchestral passages during climax

Upsampling Algorithm Comparison

An unexpected finding was how differently each DAC responded to upsampling algorithms:

SoX Upsampler:

  • Provided clearly best results with Benchmark DAC3
  • Allowed fine-tuning of multiple parameters
  • Optimized settings discovered in previous review worked perfectly with Benchmark

r8brain Upsampler:

  • Improved bass with Yamaha RX-A8A, but I still preferred my SoX settings overall
  • Added more warmth and bass presence to Yamaha’s presentation
  • Limited adjustment parameters compared to SoX
  • Default settings provided reasonably good results with Yamaha at 384kHz (multiple of 2 upsampling)

Technical Analysis

What explains these consistent differences between two high-quality DAC implementations? Several factors likely contribute:

1. Power Supply Implementation: The Benchmark DAC3 benefits from a dedicated, optimized power supply, while the Yamaha’s DAC section shares power resources with its many other functions. Clean power has proven crucial in my previous optimizations.

2. Jitter Reduction Chain: My reference system includes multiple stages of clock regeneration (UltraRendu → IsoRegen), significantly reducing timing errors that affect soundstage reproduction and bass definition.

3. Analog Output Stage Design: The Benchmark’s analog output stage is designed specifically for two-channel audio, while the Yamaha must accommodate multichannel processing and other features.

4. Signal Path Optimization: The Yamaha appears to have different internal signal paths for PCM and DSD (evidenced by the relay click when switching formats), with the PCM path likely more direct and optimized.

5. Implementation Priorities: The Benchmark focuses entirely on audio quality, while the Yamaha balances this with versatility, surround and DSP features, and other design considerations.

Conclusion

Does the Yamaha RX-A8A with its ES9026PRO DACs outperform my optimized UltraRendu → Benchmark DAC3 chain? In short: no. Despite impressive specifications and capabilities, the Yamaha’s DAC implementation cannot match the emotional connection, soundstage depth, and natural presentation of my reference system.

However, this doesn’t diminish the Yamaha’s accomplishments. As a multi-purpose receiver, its performance is impressive, especially with PCM 384kHz and Audirvana upsampling. For many listeners, particularly those without extensively optimized systems, the Yamaha could provide outstanding results.

Key takeaways for Audirvana users:

  1. Implementation matters more than specifications - The Benchmark DAC3 with “only” DSD64 capability consistently outperformed the Yamaha’s DSD256 and PCM384 capabilities because of superior implementation.
  2. Different DACs prefer different formats - The Yamaha performed best with high-resolution PCM, while the Benchmark excelled with DSD64 conversion.
  3. Upsampling algorithm selection matters - SoX worked best with the Benchmark and was also my favorite with the Yamaha, but r8brain better complemented the Yamaha’s neutral bass character (my SoX settings sounded leaner with the Yamaha’s DAC).
  4. Power supply quality remains crucial - The benefits of my linear power supplies were evident throughout testing, reinforcing their importance in digital audio chains.
  5. System synergy trumps individual components - The cohesive design of my reference system created a performance level greater than the sum of its parts.

This comparison reinforces what many audiophiles discover: specifications alone don’t tell the whole story. My multi-box approach with the Benchmark might seem “overly complicated” compared to the Yamaha’s newer technology, but careful implementation, optimization, and system synergy created a musical experience that the integrated solution couldn’t match. This obviously also comes at a significantly increased cost. But I think Audirvana users aim for maximum musical enjoyment and the UltraRendu isn’t the most expensive solution, even with a good linear power supply.

For those with Yamaha receivers wanting to maximize performance with Audirvana, I recommend:

  • Use PCM 384kHz/32-bit rather than DSD conversion
  • Try both the SoX and r8brain upsamplers
  • Consider a clean power supply for your streaming device
  • Definitely replace the receiver’s bundled power cord. Even an inexpensive “audiophile” cable from Amazon or Temu around €60-€90 can do a lot (but be sure to try them out because some of them, especially silver infused ones, resulted in worse sound for me — I found neutral shielded copper best as an inexpensive step up. Although I ended up in higher end price ranges from Canadian and U.S. suppliers for my amps and UltraRendu’s LPS. Make sure that your power cables are at least 1,5m/5ft in length, 6ft is ideal).

For now, my optimized Audirvana → UltraRendu → IsoRegen → Benchmark DAC3 chain remains my personal reference for “magical” musical enjoyment.

System Details

Equipment Used:

  • Ethernet Switch: Uptone Audio Etherregen, powered by Uptone JS-2 Linear Power Supply at 12VDC
  • Player: Mac mini (late 2012) with Uptone Audio DC-Conversion / Linear Fan Controller Kit, powered by Audiophonics LPSU200 LPS
  • Network Streamer: Sonore UltraRendu powered by Uptone JS-2 LPS
  • USB Reclocker: Uptone IsoRegen powered by sBooster LPS (set to 6.5V) [corrected – in the initial version of my review I incorrectly wrote that the IsoRegen is powered by the JS-2 LPS] and connected to the DAC using an Uptone USPCB adapter
  • DAC: Benchmark DAC3
  • Receiver/Alternative DAC: Yamaha RX-A8A
  • Speakers: Bowers & Wilkins 802 D2

Optimized Audirvana Settings (Reference):

  • SoX Filter Bandwidth (% Nyquist): 94.5
  • SoX Filter Max. Length: 23296
  • SoX Filter Anti-Aliasing (%): 96
  • SoX Filter Phase (Min. Phase to Linear): 66
  • DSD Sigma-Delta Modulator Filter Type: A (4th order)
  • Safe volume reduction before DSD upsampling: -4dB

Best Yamaha Settings:

  • Format: PCM 384kHz/32-bit
  • Upsampler: SoX or r8brain (both need tweaked settings, r8brain seems to result in slightly better bass response)
  • Yamaha Pure Direct mode enabled

—-

Note on AI assistance

I used Claude.ai to help organize my findings from this extensive listening comparison. During approximately eight hours of testing, I typed about 7,000 words of raw observations and questions into Claude, including detailed listening notes for all seven reference tracks. Claude already had knowledge of my system setup and previous optimization results (from my earlier Audirvana SoX tweaking review), which provided helpful context. Claude helped transform the information into this more structured 2,000-word review, where it identified three representative tracks where the differences were most revealing and confirmed over multiple listens. The analysis entirely reflects my actual listening experience and conclusions, with Claude helping with organization and technical explanations for what I was hearing. I made final edits and refinements before posting.

2

This is a nice AI composed piece… However the ‘test’ is flawed and dripping with subjective cognitive-bias… See my response here:

3

I use AI to make my work more readable and organized. I can assure you that the opinions expressed and audition results are entirely mine. I have the full logs of my recent auditions stored in a project. I have found it hugely helpful. I tried pasting the full audition transcript below but it seems the size of this reply is limited.

Of course an audition like this is subjective. It is also honest.

I’m surprised that you seem to imply that an integrated receiver should sound equally good as a dedicated stereo DAC fed by an LPS-powered streamer. This is my fourth Yamaha receiver and none of them sounded as good for stereo music (including streaming directly from TIDAL and, since last year, Qobuz) as the separate DACs I’ve owned. You may call it confirmation bias all you want, my experience obviously differs from yours.

2 Likes
4

I like to think of digital to audio conversion as divided into several different stages. There are the input stage (including the clocking), the conversion stage, and the output stage (including the final hardware analog filter and output circuitry).

Audirvana can under some circumstances help with the input stage (integer mode), can help a lot with the conversion stage (upsampling and sigma-delta modulation), and a bit with the output stage under some circumstances (with the signal it sends to the final analog filter in DACs that don’t impose their own digital filtering).

But that still leaves a lot up to the hardware implementation in the DAC - the input circuitry and clock, the final hardware analog filter and output circuitry. That hardware implementation can easily have more to do with the final sound than any changes that Audirvana can make. Audirvana can absolutely in my subjective experience improve the sound of a DAC, and some more than others. But the result of a comparison between two DACs will be very much dependent on hardware design as well as what can be done with Audirvana.

4 Likes
5

What do you think is the fundamental DAC topology difference? They both use Hyperstream II output to the D/A circuits… I think the difference is the DAC-3 implements four (4) ES9028PRO chipsets and the Yamaha implements two (2) ES9026PRO chipsets…

They have designed the DAC 3 for PCM production work up to 24/192kHz… The ES9028PRO chipset is capable of higher PCM sample-rates to 768kHz and DSD1024 sample-rates, but they have chosen to throttle it in this design… It’s good for the market it is targeted to support.

Given the fundamental DAC implementation differences (significant) your ‘Technical Analysis’ is conjecture for the most part, and should be presented as such… not with unsubstantiated opinion and hubristic assertions as you have propagated here… I’m not defending the Yamaha design… as an audio engineer audiophile musician and electronics design engineer in R&D with over 40 years working in digital production, I am questioning your authority to purvey such subjective observational opinion garbage here in this forum about the hardware implementations… For your information, I installed the first 8-track ProTools system in the world… You can bet there are many Audirvāna user systems that are fully capable of blowing away the Benchmark system you are so proud of… Most of us don’t care what you have put together for a system… What matters is what you feel in the final audition… no matter the amalgamation of the system components… It’s all very subjective and there is no need to clutter things-up with a bunch of subjective opinion that we have no way of corroborating.

:notes: :eye: :headphones: :eye: :notes:

6

Sorry for nitpicking but there are too much variables in this comparison.
A better comparison would be to have the same upstream set-up and feeding both the Benchmark and the Yamaha via their USB inputs. So the upstream set-up would be identical and you could directly compare the DACs.

7

I believe the receiver’s USB input does not work for USB audio. It is designed to attach a storage medium with files. Correct me if I’m wrong, because I would have loved for my receiver to also be a USB DAC. But as far as I know I cannot connect the UltraRendu to it.

I am very pleased with the Yamaha’s performance in surround, including with Blu-ray Audio and DSD from multichannel SACDs coming over HDMI from my Pioneer UHD player. I am surprised that DSD over UPnP into the Yamaha didn’t impress me more than PCM. I suppose I should audition a DSD file on a USB stick to compare. And I should re-listen to the MoFi SACD of Love Over Gold and compare it to the upsampled PCM. After my latest Audirvana SoX optimizations the UltraRendu going into the Benchmark DAC sounded impressively similar to the SACD played on the Yamaha.

8

OK, maybe if it doesn’t have an USB-B input.
Then I would suggest this comparison:
Feeding the Benchmark directly via the USB output of the Mac vs feeding the Yamaha via UPnP.
So Ultrarendu and IsoRegen are removed and provide a better comparison of both DACs.

9

That’s a good point, but my comparison is explicitly based on my current reference system. In earlier tests the UltraRendu already proved to improve the sound of my system. I would not remove UltraRendu nor IsoRegen from my Benchmark DAC3 setup.

I’m 100% sure that the Benchmark straight from the Mac mini’s USB port sounds less magical. Whether it still beats the Yamaha receiver would require another careful audition.

My goal for this test was to see whether my existing reference setup for stereo listening can be matched by my existing Yamaha receiver. If I had known a way to make my previous Yamaha receiver (the RX-A2070) sound as good as an external DAC with Audirvana (then Aune, now Benchmark) then I would have avoided the cost of the external DACs altogether. Every component in my system was bought because it improved the music… not because I like to spend money :slight_smile:

1 Like
10

@mhsmit

When I used multichannel for movies, I had the Yamaha RX-A3060.
I tested it at the time. It’s not bad for listening to music, but it was never better than a dedicated DAC.
My Yamaha played DSD files better than PCM.
Today I only have a stereo system using a Mac mini M2 with Audirvana and an IFi Neo iDSD 2 DAC, which I hope to keep for a long time.
What I didn’t like about the Yamaha was hearing the relays every time I switched from PCM to DSD.

2 Likes
11

You should reconsider this approach, since your (your?) postings’ »soul« seems dead somehow …

12

Exactly why we don’t need the interpretive rhetoric of a subjective assessment that we have no way of corroborating… It is valueless in this forum.
:notes: :eye: :headphones: :eye: :notes:

13

I find that sincerely interesting — it is obviously not my goal for my posts to sound dead :sweat_smile:. Can you give me an example of what aspect of the review text makes you feel that way? What annoys you the most?

For background:

My full input to Claude during the testing yesterday was about 7000 words, of which about half were raw audition notes written while the reference tracks were playing. Claude also has the notes from my Audirvāna SoX tweaks, and knows my exact system setup. It also has notes I made during recent power supply and cable upgrades.

At my request Claude turned it into the draft for the above review of about 2000 words. I was actually impressed by how well the review draft reflected my memory. I made some final minor textual changes and additions (such as the detailed recommendations on the power cable) before posting.

It is definitely true that Claude filtered out some spontaneous nuances. And it came up with the observations for the three reference tracks (out of seven) where it noticed my comments had the most consistency over multiple listens.

Again, I’m really interested in hearing what annoyed you. I use AI mostly to organize my own thoughts and research. This is the first time I’ve used it for documenting experiences with my private audiophile system. It definitely has helped me. I certainly don’t need to post it anywhere, and if I do, I want it to be useful and certainly not sound “dead.”

From now on, I’m definitely going to include how AI was used. I’ll be happy to also post the full listening notes if that is of interest.

14

:notes: Re-Humanize Yourself…:notes: Stop feeding the monster…

1 Like
15

I respectfully disagree. I find AI an indispensable tool. But I do think there is a need for strict “AI etiquette”. From now on I’ll always add a “how I used AI” to the end of my work. I just added the following to the end of the review:

Note on AI assistance

I used Claude to help organize my findings from this extensive listening comparison. Duringji approximately eight hours of testing, I typed about 7,000 words of raw observations and questions into Claude, including detailed listening notes for all seven reference tracks. Claude already had knowledge of my system setup and previous optimization results (from my earlier Audirvana SoX tweaking review), which provided helpful context. Claude helped transform these information into this more structured 2,000-word review, where it identified three representative tracks where the differences were most revealing and confirmed over multiple listens. The analysis entirely reflects my actual listening experienced and conclusions, with Claude helping with organization and technical explanations for what I was hearing. I made final edits and refinements before posting.

16

Then you don’t need to qualify anything… :sunglasses:
Let’s enjoy the existential experience for as long as we can… no need to accelerate our demise… :roll_eyes:

17

I hate to think our postings here in an audio forum would be critiqued as if they were composition lessons. Do what works for you.

3 Likes
18

Haha agreed. There’s no way AI is going to replicate our personal enjoyment of music. I find it cool that human hearing and audiophile reproduction are areas that cannot be fully explained by science and math, just like love is, or the appreciation of good wine.

19

Be careful for what you want and what you teach…

20

It will define our personal enjoyment through bias… Stop feeding the monster… Ever notice how modern vehicles of different genre look very similar? Rationalization through computerization, tends to homogenize the world around it… For your information… I grew up in what is known today as Silicon Valley… Alphabet (formerly Google) headquarters sits on a garbage dump that contains some of my childhood stuff… I went to the rival high school of Steve Jobs and Steve Wosniak… Steve J. would be the same age a me… We had the same baby-sitter when we were youngsters… :wink: