Topping D900 DAC Review

[Endibol]

With a sine sweep you can see tiny variations in frequency response, perfect for assessing 20Hz to 20kHz flat-ness.

With noise you can assess the out-of-band filter performance down to -100dBFS and beyond.

Different test signals for assessing different aspects of frequency response, but in the end they both show the same attenuation (just displayed at different scales).

@staticV3 Thank you for the clarification.

 

[HiFiHelp]

View attachment 517843

1.6 uVrms noise and 13 Vrms output equates to 138 dB of dynamic range, but Topping specs the device at 131 dB of dynamic range. Your measurements also indicate that it has a max dynamic range of 131 dB. Is the noise floor spec inaccurate?

 

[antcollinet]

1.6 uVrms noise and 13 Vrms output equates to 138 dB of dynamic range, but Topping specs the device at 131 dB of dynamic range. Your measurements also indicate that it has a max dynamic range of 131 dB. Is the noise floor spec inaccurate?

Look at the output voltage used for the test at the bottom of the table you quote. 5V, not 13V - an 8dB difference.


Amir's test used 4V

 

[HiFiHelp]

Look at the output voltage used for the test at the bottom of the table you quote. 5V, not 13V - an 8dB difference.


Amir's test used 4V

Amir's 4.11 Vrms measurement shows only 120 dB of dynamic range. 13 Vrms is +10 dB higher than 4.11 Vrms, meaning that if the noise floor stays the same, the dynamic range could only be increased to 130 dB at 13 Vrms output.

 

[antcollinet]

Amir's 4.11 Vrms measurement shows only 120 dB of dynamic range. 13 Vrms is +10 dB higher than 4.11 Vrms, meaning that if the noise floor stays the same, the dynamic range could only be increased to 130 dB at 13 Vrms output.

Again - Toppings own measurements were carried out at 5V

Further, do you think the noise floor will stay the same as you increase the gain on the output? Seems unlikely to me.

Also - at these levels of noise we are approaching the limits of the analyser's capability to measure them. Exceptional care is needed with test set up to even achieve that level of accuracy. You are in the realm of arguing "how many angels can dance on the head of a pin" We are way beyond the limits of audibility here.

 

[HiFiHelp]

Again - Toppings own measurements were carried out at 5V

Further, do you think the noise floor will stay the same as you increase the gain on the output? Seems unlikely to me.

Also - at these levels of noise we are approaching the limits of the analyser's capability to measure them. Exceptional care is needed with test set up to even achieve that level of accuracy. You are in the realm of arguing "how many angels can dance on the head of a pin" We are way beyond the limits of audibility here.

I see. I assumed that the 13 Vrms output was with its <1.6 uVrms noise level spec. Thank you for the clarification. The D900 seems like a complete waste of money when the D90 iii and D70 Pro Sabre achieve much higher dynamic range figures.

 

[antcollinet]

I see. I assumed that the 13 Vrms output was with its <1.6 uVrms noise level spec. Thank you for the clarification. The D900 seems like a complete waste of money when the D90 iii and D70 Pro Sabre achieve much higher dynamic range figures.

Really? D90III specced at 130 - essentially same as D900. Amir's measurements similar also.

Plus again - when you are so far beyond audibility, why do you care? Compare them on price for features for quality. There is no audible difference.

 

[HiFiHelp]

Really? D90III specced at 130 - essentially same as D900. Amir's measurements similar also.

Plus again - when you are so far beyond audibility, why do you care? Compare them on price for features for quality. There is no audible difference.

D90 iii Sabre, not discrete. There is multi-channel content that requires -9.5 dB preamp to prevent clipping when downmixed to stereo, including the LFE channel at +7 dB gain into the L/R channels independently (for +10 dB mono gain per LFE standard). On top of a high negative preamp EQ, the dynamic range requirements of a system for no audible noise can be really high.

 

[antcollinet]

D90 iii Sabre, not discrete. There is multi-channel content that requires -9.5 dB preamp to prevent clipping when downmixed to stereo, including the LFE channel at +7 dB gain into the L/R channels independently (for +10 dB mono gain per LFE standard). On top of a high negative preamp EQ, the dynamic range requirements of a system for no audible noise can be really high.
View attachment 537194

Consider - even with only 100dB dynamic range, you have to be listening at 100dB to get the noise up to the threshold of audibity. Except even then it isn't, since the noise is wide band - only a small part of the noise power will be in the sensitive range our hearing at around 1 to 5 kHz. And that is for people with perfect hearing - perhaps a few teenagers.

And then this level of DAC noise is likely going to be drowned in that coming from your amplification, which is going to be at least 10 times higher (even if you are using the lowest noise amps measured here) - most likely 100 times higher if not

In real world listening (typical room, etc) you are not close to having a problem with these devices.

 

[HiFiHelp]

Consider - even with only 100dB dynamic range, you have to be listening at 100dB to get the noise up to the threshold of audibity. Except even then it isn't, since the noise is wide band - only a small part of the noise power will be in the sensitive range our hearing at around 1 to 5 kHz. And that is for people with perfect hearing - perhaps a few teenagers.

If you do not have analog volume control, the noise floor of the DAC is generally fixed regardless of the volume level. And at lower attenuation levels for really high dynamic range content playback, most of the time the content is operating at far below full scale (dialogue in movies, for example). My Topping A900 has a max dynamic range of 151 dB. Why would I pay more for a DAC with objectively worse performance?

Also, old age doesn't cause hearing loss: https://acoustics.org/hearing-loss-in-old-age-isnt-due-to-normal-aging/

 

[antcollinet]

he noise floor of the DAC is generally fixed regardless of the volume level.

If the noise floor is inaudible in the silence between tracks at your listening volume - it remains inaudible no matter how far you turn the DAC volume down. As you say - it is fixed.

 

[HiFiHelp]

If the noise floor is inaudible in the silence between tracks at your listening volume - it remains inaudible no matter how far you turn the DAC volume down. As you say - it is fixed.

When I am talking about hearing noise, I am using noise-isolating in-ear headphones. I'm not saying 1.5 uVrms noise is very high (90.5 dB 50 mV SNR), but why in the world would you pay so much for that performance when D90 iii Sabre has 1.1 uVrms noise (93 dB 50 mV) for the same output voltage?

 

[antcollinet]

Topping A900 has a max dynamic range of 151 dB

A weighted.

Why would I pay more for a DAC with objectively worse performance?

Because you don't need that performance - and an alternate DAC might have other features you are willing to pay for. If that doesnt apply to you, then fine - but it is a personal decision, not a universal one.

 

[nanook]

Just to get the dimensions we're talking about transparent :

I'm using a D-6s and a pair of old Genelecs 1030A.
The Genelecs are set to minimum sensitivity, the WiiM Pro (feeding the DAC) is configured to volume-limit = 80% => -12dB (I use boost room PEQs and the volume at the minimum volume setting of the WiiM streamer was still too loud for background listening).
This may end up with playing music having peaks at 50 or even down to 62dB below FS of the DAC.

123dB (D-6s SINAD @5Vrms) - 12dB (volume-limit) - 50dB (min volume at WiiM) = 61dB dynamic range at minimum volume setting at DAC output.
My max. listening level is round around 100dB SPL at max volume (Genelec: 100dB SPL, 1m @ +6dBU; 2 speakers so +3dB, distance is probably 2.5m, so we end up slightly below 100dB SPL)

100dB SPL - 50dB = 50dB SPL at min. volume
This is almost background listening.
Still I do have a dynamic range of 123dB - 62dB = 61dB when looking at the DAC performance. This is still good for ca. 11dB below hearing threshold.
Even in some quiet place in the countryside you're fine

PS: In case I messed up the calculations, please correct me.

EDIT: corrected the numbers

 

[Max_fl]

Received mine recently, it sounds better than D90 III Discrete, I hear more details in my music. Overall, it's a good upgrade for me.

 

[Jimster480]

Received mine recently, it sounds better than D90 III Discrete, I hear more details in my music. Overall, it's a good upgrade for me.

There's no way that any of that is possible unless your other device is broken. These devices are all state of the art and as such they are beyond the threshold of human hearing in terms of their performance.

 

[DrCWO]

There's no way that any of that is possible unless your other device is broken. These devices are all state of the art and as such they are beyond the threshold of human hearing in terms of their performance.

From technical point of view you are absolutely right.

What @Max_fl did was a sighted listening test. As Sean Olive showed here the perception is different if you do it sighted. So please don't blame him for hearing more details. This is his REAL perception.

But for sure performing a blind ABX test you will not be able to distinguish the SOTA DACs

 

[Max_fl]

There's no way that any of that is possible unless your other device is broken. These devices are all state of the art and as such they are beyond the threshold of human hearing in terms of their performance.

As you say
Although, whoever wants to buy my "broken" D90 III Discrete is welcome to DM.

 

[Tokyo_John]

These SOTA DACs are saturated in terms of transparency and noise. I think the only audible differences one can get from DACs having SINAD well beyond human ability to parse is due to factors like:
(1) Rolloff filter variance...and only if one has younger ears that are sensitive to rolloff frequencies.
(2) Non-linear interactions between DAC characteristics (including rolloff) with other components in the stream (amps, transducers, etc.).
(3) Flaws in system setup (e.g., improper grounding) that affect certain DACs but not others.
(4) Manufacturing defect(s) in a particular device that cause it to perform below its designed level.

The probability that these will manifest for a particular user/system is small in modern times, though non-zero. The first issue only arises if the DAC uses a non-standard rolloff, and can usually be mitigated by having a broad selection of rolloff filters that are well-executed by the DAC, (many DACs do offer this capability). The second is a problem related to other components, which can also be mitigated by swapping them out, etc..

Here is the thing that makes good business sense for Topping to introduce products like the D900: Much of the hi-fi audio market is still irrational, in spite of the hard work that Amir and others here at ASR have been doing for some years. Some people still spend ungodly sums of money on pure snake oil. There may also be other biases to overcome in the market, such as people who are innately skeptical about "cheaper" Chinese products and would not feel good about including them in a system that they are otherwise investing a lot of money to build...they'll never get over seeing a sub-$1000 Chinese DAC in their system, it doesn't matter how well it performs in analytical tests. Topping can indeed address this prestige issue by offering fancier gimmicky DACs at higher price points. There is a certain psychology at work, in which sometimes simply raising the price of a product line generates more demand because it is perceived as being more valuable (yes, it is irrational, but this is how certain market segments operate).

For the more rational side of markets, the best thing that DAC manufacturers can focus upon is product functionality, QC, reliability, support, and price. All but one of these are still engineering challenges. Support is vital for a company's reputation, and under-performing in this area will negate any gains from trying to sell in the more luxury side of the market to raise prestige.

 

[antcollinet]

This is his REAL perception.

True - but based on unconscious manipulation of the sound in the brain. Nothing to do with the sound from the dac reaching his ears.

And it is NOT consistent, being based only on novelty of new gear. There is no real benefit. After days or weeks the sound will still be the same - and the next time he hear's somthing new - there will be new non existent improvements to be heard.