Topping Centaurus R2R DAC Review

[peng]

On a different but somewhat related topic (if not, I hope our mod will delete it for me), I wonder if there exist a device that could easy add a variety of harmonics to get the effects that antcollinet described in post#38? NP, iirc had a sort of DIY, prototype one but I really want to buy one that is a finished off the shelf product that can be readily wired into some point of my preamp/dac/dsp signal chain, and the harmonics addition should be reasonably variable and the adjustments visible on its own display or at least a PC. If such a device doe not exist, I hope it will, in the near future.

It seems to me frequency response and THD, and may be its frequency dependence too, are probably the main reason for audibly and obviously measurably different sound quality, as other metrics such as linearity, FR, DF, IMD, DR/SNR, multitone inputs, (usually very low if THD is very low), cross talk, and all the stuff that Amir has been measuring tend to be transparent enough for most devices that are on his recommended list.

I think this is sort of still on topic because there must be members who are interested in why R2R DACs sound different, assuming some do sound different in blind listening tests.

 

[unpluggged]

where dCS land

They land in a landfill.

 

[Scytales]

How does R2R approaches differ from something like the Marantz SA-10 with its discrete DAC approach? It seems like this is just mechanical watchmaking for audio?

No service manual seems to be available. Only low-res photos are to be found of the internal, without much visible details.

From what I am able to see, it may be that the Marantz uses a kind of conversion method akin to the Sygnalist DSC-1 discrete D/A converters : everything gets converted to a very high sample rate (11.2 MHz from what I read in some reviews) 1 bit signal that is passed through a moving average low-pass filter.

 

[staticV3]

It would be interesting to know where dCS land in terms of performance in comparison.

dCS Bartok (with headphone amp) Measurements - GoldenSound

The dCS Bartok is a 'summit fi' level DAC with a hefty pricetag of £12,500. But whilst the performance is great in many areas, others leave some questions unanswered.

goldensound.audio

And, whaddya know:

dCS is threatening me with a lawsuit over a negative review [Update Inside] - GoldenSound

dCS has threatened me with a "7-figure lawsuit" over my coverage of the dCS Bartok in 2021. This video discusses my response to their claims

goldensound.audio

 

[antcollinet]

On a different but somewhat related topic (if not, I hope our mod will delete it for me), I wonder if there exist a device that could easy add a variety of harmonics to get the effects that antcollinet described in post#38? NP, iirc had a sort of DIY, prototype one but I really want to buy one that is a finished off the shelf product that can be readily wired into some point of my preamp/dac/dsp signal chain, and the harmonics addition should be reasonably variable and the adjustments visible on its own display or at least a PC. If such a device doe not exist, I hope it will, in the near future.

It seems to me frequency response and THD, and may be its frequency dependence too, are probably the main reason for audibly and obviously measurably different sound quality, as other metrics such as linearity, FR, DF, IMD, DR/SNR, multitone inputs, (usually very low if THD is very low), cross talk, and all the stuff that Amir has been measuring tend to be transparent enough for most devices that are on his recommended list.

I think this is sort of still on topic because there must be members who are interested in why R2R DACs sound different, assuming some do sound different in blind listening tests.

I'm not aware of any devices, but if you want to play around with distortion to see what it sounds like, then @pkane 's "distort" can do it on a PC.

If just interested in harmonic distortion (and if you are looking for "pleasing" ditortion, you probably are) then his pkharmonic plugin might be more interesting.

 

[Mikig]

bravo Topping. A great device. At this point with measures of this kind, even R2R becomes a highly justified purchase. So, if you like the particular and are attracted by the fascinating engineering complications you know that today with devices of this kind you lose practically nothing on an objective level. It is nice to see the giant steps that these manufacturers have made in a handful of years. Let's face it: technology and innovations now come only from them... the noble brands of hi-fi should start asking themselves some questions...:

Thanks Amirm for the test!!!

 

[meracus]

Serious question: how do we think that R2R DACs got their audiophile status?

Elitism … what I have is rare , what you have is found in a food processor type of attitude
…. It’s so costly to make that most manufacturer gave up on the idea … etc etc

 

[welwynnick]

If only holo audio cyan 2 get measured too...

Well, what do you know - it's not quite as good as the Centaurus. The Cyan 2 is rather better though, and is probably similar to the Topping.

Measurements of HoloAudio Cyan DAC&HeadphoneAmp - L7Audiolab

台式设备 众所周知，我挺欣赏HoloAudio的设备，就实来说，我和HoloAudio的开发者Jeff关系不错。但是，他一直有一条产品线我没有测量过-----青（解码耳放一体机）。今天要发布的就是网友送测的这台机器

www.l7audiolab.com

 

[unpluggged]

even R2R becomes a highly justified purchase

What makes it justified against a cheaper and unflawed DS DAC?

if you like the particular and are attracted by the fascinating engineering complications

I'm attracted by real engineering merits. For instance, by products that are built to last and have a long lifecycle and adequate aftersales support processes. By manufacturers who make an effort to provide comprehensive documentation. By products that have really useful features instead of gimmicks and that are not only not subject to planned obsolescence, but kept up to date long after the initial release. This is what makes "fascinating engineering" in my book.

giant steps

Yes, for them these are indeed giant steps.

Let's face it: technology and innovations now come only from them...

From people who only recently learned to read IC implementation notes and discovered the existence of the AP555 analyzer? From those who are struggling to produce an amplifier that does not kill your headphones? Good joke.

 

[welwynnick]

On a different but somewhat related topic (if not, I hope our mod will delete it for me), I wonder if there exist a device that could easy add a variety of harmonics to get the effects that antcollinet described in post#38? NP, iirc had a sort of DIY, prototype one but I really want to buy one that is a finished off the shelf product that can be readily wired into some point of my preamp/dac/dsp signal chain, and the harmonics addition should be reasonably variable and the adjustments visible on its own display or at least a PC. If such a device doe not exist, I hope it will, in the near future.

Topping and Okto make DACs that have "adjustable" 2nd & 3rd order harmonic distortion.
I think it's user control of the ESS distortion compensation functions on Registers THD_C2_0 through HD_C2_7 and THD_C3_0 through THD_C3_7: THD_C2_0.

A 16-bit signed coefficient for correcting for the CH1 second harmonic distortion.
output = x+c2*x2+c3*x3 ...etc

Topping D90LE - "Valve" and "Transistor" effects measurements

Hi I've performed a few quick measurements with the Topping D90LE DAC 5V at Max level (USB in, 48kHz, Right Channel) (ADC: E1DA ADC 4.5V Mono, measured with Virtins Multi Instrument 3.9.6) "Standard" mode In "Valve" mode We see a big 2nd harmonic peak (0.32%), with a more limited 3rd...

www.audiosciencereview.com

Okto Research

Objective performance

www.oktoresearch.com

Advanced submenu
DPLL BW AES
THD h2
THD h3
8 0 0 - DPLL BW AES: controls bandwidth of the de-jittering mechanism for AES/EBU inputs. Increasing the value will allow for operation with more jittery signal sources at a cost of slightly reduced jitter suppression - range: 0 to 15 (default: 8) -
THD h2: introduces a 2nd order nonlinearity to the DAC's transfer curve, potentially neutralizing second harmonic distortion generated by the DAC chip or subsequent analog circuitry. This setting is intended for technically proficient users with access to a high-end audio analyzer. - range: -99 to 99 (default: 0) -
THD h3: same as the previous setting, but for 3rd order harmonic distortion - range: -99 to 99 (default: 0)

 

[peng]

I'm not aware of any devices, but if you want to play around with distortion to see what it sounds like, then @pkane 's "distort" can do it on a PC.

If just interested in harmonic distortion (and if you are looking for "pleasing" ditortion, you probably are) then his pkharmonic plugin might be more interesting.

Excellent! I did not know there is such a plugin, many thanks to pkane.

 

[totti1965]

Where is the f…ing difference between the Topping D90 III Discrete (aviable scince August 2024) and this device?

 

[JDS]

This is a review, detailed measurements and listening test of the Topping Centaurus R2R stereo balanced DAC with PEQ filter support. It was sent to me by the company and costs US $999.
View attachment 413793
The unit has the elegant look of the new Topping DACs with high resolution, high contrast screen, nicely highlighting the volume level. It is fully graphic and can be changed to show spectrum, VU, etc. The included Bluetooth remote felt like it had a slight lag to it. Nothing serious but I don't remember this before. There is no rotary volume control. I use the remote for home use but on desktop, a rotary control is nicer for quick adjustments. Back panel shows the nice feature set we have grown to like from Topping:
View attachment 413794
Of note, we have the very useful trigger in/out for automatic turn on/off of downstream amplifier.

As the heading indicates, this is Topping's first discrete "R2R" DAC instead of using an integrated DAC solution. R2R performance can range from close to awful to respectable -- but never as good as IC based solutions. Let's see where the Centaurus lands. All tests are in default 4 volt mode instead of 5 volts (and "PRE" mode).

If you are not familiar with my DAC measurements, please watch this video first:

Topping Centaurus R2R DAC Measurements
As usual we start with our dashboard with XLR output:
View attachment 413795
This is better than I expected! Topping says THD+N is better than 0.0005% and we are getting half of that, resulting in the Centaurus landing in our "Excellent" category:
View attachment 413796
View attachment 413797

Harmonic distortion is higher than state of the art but all peaks are below -120 dB, making them inaudible.

RCA performance is naturally a bit lower:
View attachment 413800

Noise performance is essentially state of the art:
View attachment 413801

Jitter test output is not as squeaky clean as we normally see from Topping but is transparent:
View attachment 413802

Distortion was kept in check with our punishing 50 Hz tone into 600 ohm load:
View attachment 413807

Where we go sideways with R2R is in mid levels:
View attachment 413803

We have about 20 to 25 dB more distortion and noise at -33 dB than state of the art DACs. The quietness that topping brings to the table shows up at the start of the sweep where distortion is still very low.

Linearity however, has taken the smallest, negligible hit:
View attachment 413804

I was disappointed that none of the filter settings resulted in flat frequency response to 20 kHz:
View attachment 413805
View attachment 413806
On the other hand R2R fans get full menu of what they are asking with myriad of filter settings.

With default filter, there is still some out of band noise to reduce performance but impact is low:
View attachment 413808
Go to "non-oversampling" mode and ultrasonic noise shoots through the noise, causing that off-the-charts green line. Best not to have a tweeter that attempts to go above 20 kHz!

Topping Centaurus R2R DAC Listening Tests
I gave the DAC a quick test to see if I can the magic of R2R DAC. Connected the XLR output to Topping A90 headphone amp and drove the Dan Clark E3 headphone with it. I Was listening to one of my favorite tracks from Civil Wars, the remake of Billie Jean on the RME ADI-2 Pro. After quickly moved the headphone jack to Centaurus A90 chain, I noticed the highs collapse a bit but raspiness of the Jean-Paul White became more pronounced and pleasing! Was quite a surprise until I plug the headphone back into the RME, cranked up the volume to similar level and it too produced the same sound. Teaches me not to do sighted tests like this.

Conclusions
Objective performance of Centaurus seems to be the best that can be done with R2R technology. Topping brings to the table ultra low noise, equalization and great packaging, offsetting some of the shortcomings of that technology. Ultimately it can only polish that so much. Fortunately, distortion levels are likely still well below audibility, sans the rolled off highs that may be audible to younger folks. They should keep in mind as they change filters that the roll off changes with it.

Once a month someone asks me what R2R DAC they should buy. They say they have tried all the ESS/AKM DACs and want to play with something new. When I tell them I have nothing to recommend, some accept that, but others downright get angry with me! Now I can tell them to go and get the Centaurus and sleep easy that it at least doesn't compromise much fidelity.

While I would not buy or use an R2R DAC, I am still going to recommend the Topping Centaurus for those who want an R2R technoloy.
------------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Getting an R2R to measure this well is a real accomplishment from an engineering standpoint. But I don't understand the marketing choice here. I assume there isn't a lot of overlap between tweako subjective audiophiles and Topping customers, who can get more performant DACs from Topping's own catalog for much less. So who is this for?

 

[Purité Audio]

It will appeal to those who believe , despite evidence of measurement that there is some ‘magic’ in resistor to resistor designs.
Keith

 

[Billy Budapest]

Voted “fine” because although most measurements are good and faults are below the threshold of audibility, I do not see the point of this architecture.

 

[Billy Budapest]

Getting an R2R to measure this well is a real accomplishment from an engineering standpoint. But I don't understand the marketing choice here. I assume there isn't a lot of overlap between tweako subjective audiophiles and Topping customers, who can get more performant DACs from Topping's own catalog for much less. So who is this for?

Exactly.

 

[antcollinet]

Where is the f…ing difference between the Topping D90 III Discrete (aviable scince August 2024) and this device?

The D90 III discrete is a 1 bit Delta Sigma Design.

This one is an R2R resistor ladder network design.

The D90 discrete being Delta Sigma probably measures better than this one (though it's not been tested here yet I think). But this DAC is good enough for those differences to be inaudble as long as a sampling theorem compliant filter is selected.

 

[YSC]

Actually for R2R the holo audio spring and may are quite good dacs (using spring 2 for years since it's release and no complaint)

 

[welwynnick]

R2R NOS DACs were never on my radar before, they were just a curiosity until the Centaurus came out. I started reading a few reviews and measurements and got curious.
The latest fave rave seems to be the Holo Audio May, which throws digital audio theory out with the bathwater in NOS mode, but many love it.

I'm struggling to understand what they're trying to do, and three things are often wrapped together, which doesn't seem right to me. Am I misunderstanding?

1. R2R DACs use a switched ladder network of filters to implement the conversion, which is an old, simple and expensive implementation.
2. Most DACs use digital filters to perform over-sampling, but NOS DACs don't do this, they convert at the native rate. This avoids pre-ringing and post-ringing.
3. NOS DACs seem to necessarily avoid using the low pass reconstruction filter, or at least one that only filters 44.1kHz

These three things are often considered collectively, but I'm sure they're separate.

1. R2R DACs are just a different architecture for performing the DA conversion. It's not very linear, and maybe people are just nostalgic? I don't care.
2. Digital filters have different frequency and time domain responses, depending on whether they're linear or minimum phase, fast or slow. The ringing is usually at 22kHz and shouldn't matter, but some people have preferences for one or another. I think NOS DACs don't use digital filters (or they're just switchable) and they don't have ringing, just lots of HF distortion instead.
3. NOS DACs don't use reconstruction filters, you just get an un-smoothed analogue output and live with that, hoping that the steps are inaudible.

John Atkinson sometimes measures the time domain effect of the different filters. Here are the Holo May in OS and NOS modes, and the MBL N31 with slow filter respectively:

The NOS time domain response to an impulse is just that - an impulse, without any pre- or post-ringing from the DAC (though the frequency domain response is compromised).
My point is that the NOS impulse response is really very similar to the MBL slow filter response above - not the same but very similar.
Is that where the benefit in a NOS DAC comes from? But can't you achieve much the same thing with a conventional DAC and a slow filter instead?

 

[totti1965]

The D90 III discrete is a 1 bit Delta Sigma Design.

This one is an R2R resistor ladder network design.

The D90 discrete being Delta Sigma probably measures better than this one (though it's not been tested here yet I think). But this DAC is good enough for those differences to be inaudble as long as a sampling theorem compliant filter is selected.

@Roland68 wrote:
The D90 III Discrete is an r2r DAC and is therefore primarily intended for a different group of buyers, such as the Gustard A26 and R26. So it is an additional option in the same price range.
The first D90 was replaced by the D90SE because AKM DAC chips were not available, the D90LE was just the additional version without MQA (at the customer's request).
In the end, Topping only released a successor to the D90SE after 3 years.

The D90 III Discrete is an r2r DAC and is therefore primarily intended for a different group of buyers, such as the Gustard A26 and R26. So it is an additional option in the same price range.
The first D90 was replaced by the D90SE because AKM DAC chips were not available, the D90LE was just the additional version without MQA (at the customer's request).
In the end, Topping only released a successor to the D90SE after 3 years.