Review and Measurements of Audio-gd NFB28.28 DAC and Headphone Amp

[andreasmaaan]

My point was multitone at different levels. Or one tone with small'harmonics. I think all natural tone consists of main frequency with small harmonics. Are all the harmics good rendered?

The result can be inferred from the IMD test + linearity test. A DAC that performs well on both will do what you are suggesting well.

 

[Bluespower]

This is actually exactly what the IMD vs level test does. IIRC correctly, Amir uses the SMPTE method which mixes a 60Hz tone with a 7kHz tone and then measures the level of the intermodulation distortion products against the output level.

This DAC performs worse on this test than almost all other DACs tested, with an optimum distortion level of about -78dB at about -17dBfs, rising to about -45db at 0dBfs.

Better DACs do -100 to -110dB IMD at -10 to 0dBfs.

Well it s not the same test to'perform levels steps by steps than to see if the dac can render a signal with all its small harmonics and other small frequencies at the same time.

 

[andreasmaaan]

Well it s not the same test to'perform levels steps by steps than to see if the dac can render a signal with all its small harmonics and other small frequencies at the same time.

There’s no theoretical reason it shouldn’t be able to, but in any case see my post #201

 

[Bluespower]

The result can be inferred from the IMD test + linearity test. A DAC that performs well on both will do what you are suggesting well.

Well why not'prove it with a test ? I,m not convinced it s the same. It's much easier to,measure well with two simple tone at same level that with a complex tone with harmonics.

 

[andreasmaaan]

Well why not'prove it with a test ? I,m not convinced it s the same. It's much easier to,measure well with two simple tone at same level that with a complex tone with harmonics.

Well personally, the idea that there would be a problem here is self-evidently false. If you doubt the well-established digital audio theory, then you're welcome to investigate it yourself, of course.

 

[Bluespower]

There’s no theoretical reason it shouldn’t be able to, but in any case see my post #201

Ok then good news and greatings from France

 

[andreasmaaan]

Ok then good news

Thanks, and sorry for the tone of my previous post

 

[RayDunzl]

Harmonics on 400Hz:

Benchmark DAC2 - inroom, so the harmonic levels aren't exactly even

NOS DAC
https://www.project-audio.com/en/product/dac-box-s-fl/

Overlaid, Benchmark left.

Generated signal. 400Hz + harmonics:

Maybe mistake on the 8th harmonic in the visuals between the two DACs. ooops.

Retest shows the 8th harmonic of the Benchmark equal to the 9th. Adjust your interpretation of the above.


-------------

Harmonics on 2000Hz

Benchmark

Pro-ject

More different at higher frequencies.

 

[amirm]

Can't you'make a waveform that mix two different tones at different levels and justs see how the dac outputs?

If you put two tones at the same frequency and phase, they just add up. They don't stay separate. It is like putting a teaspoon and tablespoon of water in a cup. They become one.

(maybe record a small sample'of accoustic guitar and compare output vs input if all hamrmonics are presents at the same level than on input).

The output is analog so subject to noise and distortions. And it also runs at a different rate than input so one to one comparison does not work. Specialized signal processing is needed to synchronize and compare them.

 

[Bluespower]

If you put two tones at the same frequency and phase, they just add up. They don't stay separate. It is like putting a teaspoon and tablespoon of water in a cup. They become one.


The output is analog so subject to noise and distortions. And it also runs at a different rate than input so one to one comparison does not work. Specialized signal processing is needed to synchronize and compare them.

If you put two different tones at different frequencies and different levels they become one more complex wave that could show differencies between dac and that would reflect more reality of sound than one tone. Is it difficult to make such a test ? Like post 208. Maybe it will show nothing but maybe there will be small differencies between chips.

 

[RayDunzl]

Maybe it will show nothing but maybe there will be small differencies

Testing in-room, which is all I can do, but corresponds to where I actually hear something, there will always be "small differences".

105 tones

Benchmark

Pro-ject DAC

240 tones

A small segment of the 240 tone waveform (in air)

 

[amirm]

If you put two different tones at different frequencies and different levels they become one more complex wave that could show differencies between dac and that would reflect more reality of sound than one tone. Is it difficult to make such a test ?

No it is not hard. That is what the IMD (intermodulation) test is. There is a strong low frequency tone and a weaker high frequency one (by a factor of 4):

In the above, the total level of those two tones is varied from -60 dB to 0 and the resulting distortion plotted.

 

[Jimster480]

TDA DAC's being described as "real and musical".... Its the wrong forum for such delusions.
This isn't headfi!

You should mail that DAC into Amir to see how it measures.
I would bet it is somewhere around 50 SINAD.

 

[Bluespower]

Ok thank you all. No i understand better the imd test and my doubts are clarified.

 

[Bluespower]

No it is not hard. That is what the IMD (intermodulation) test is. There is a strong low frequency tone and a weaker high frequency one (by a factor of 4):

In the above, the total level of those two tones is varied from -60 dB to 0 and the resulting distortion plotted.

Ok thanks. Well maybe you could experiment results of a imd test with the higher frequency much more lower than the strong one (maybe -40db lower) It would definitely cancel any doubt. It seems that the high frequency would be mixed with the noise floor in some case depending on the output level ? Maybe make it with -30db and not -40db lower high frequency. Well i found tests on my dac https://reference-audio-analyzer.pro/en/report/dac/muse-audio-4x-tda1543-nos-dac-mod.php#rw22 but mine has less thd cause i adjusted i/v resitor (i could hear the clipping)

 

[Bluespower]

Testing in-room, which is all I can do, but corresponds to where I actually hear something, there will always be "small differences".

105 tones

Benchmark

View attachment 19332

Pro-ject DAC

View attachment 19333

240 tones

View attachment 19335

View attachment 19336

A small segment of the 240 tone waveform (in air)

View attachment 19337

Very interesting test thanx a lot. For curiosity, Which one do you prefer? the nos or the benchmark?

 

[andreasmaaan]

@Bluespower another way to look at it perhaps: see in Amir’s measurements in post #1, when this DAC is fed a single 0dBfs tone, it produces about -55dB third harmonic distortion.

Imagine we now ask it to reproduce two tones, one at -3 dBfs and another at -60dBfs.

Well, we know that the third harmonic distortion of the -3 dBfs tone will be at about -58 dB. This is higher in level than the second tone we actually want it to reproduce!

There can be no hope that such a device is doing a good job in these circumstances. In fact, playing music at or near full level, anything below -55 dB will actually be buried under just the distortion from anything close to 0 dB in level.

Think about a better performing DAC now, one in which all distortion is lower in level than -100dB. Which DAC will be better at reproducing low level detail or multiple loud and quiet tones at once?

 

[Bluespower]

@Bluespower another way to look at it perhaps: see in Amir’s measurements in post #1, when this DAC is fed a single 0dBfs tone, it produces about -55dB third harmonic distortion.

Imagine we now ask it to reproduce two tones, one at -3 dBfs and another at -60dBfs.

Well, we know that the third harmonic distortion of the -3 dBfs tone will be at about -58 dB. This is higher in level than the second tone we actually want it to reproduce!

There can be no hope that such a device is doing a good job in these circumstances. In fact, playing music at or near full level, anything below -55 dB will actually be buried under just the distortion from anything close to 0 dB in level.

Think about a better performing DAC now, one in which all distortion is lower in level than -100dB. Which DAC will be better at reproducing low level detail or multiple loud and quiet tones at once?

Yes that's exactly the curves i would like to see on tests we should perform a test that show at wich level details are blurried. There is no proof that a dac with -100db noise with pure tone test will not start to blurry details at -50db also. Again it s easier for dac to have good thd with pure tone input.

 

[andreasmaaan]

Again it s easier for dac to have good thd with pure tone input.

I think you’re confusing THD with IMD. Adding multiple tones doesn’t increase THD, but it does create IMD.

And I still totally disagree with your idea that we need proof. The conclusions can be drawn with an extremely high level of confidence from the IMD, THD and linearity tests. No other tests are necessary.

Perhaps you could provide an example of a DAC that performs exceptionally on these three tests but fails the test you’re proposing?

I certainly don’t know of any.

 

[jackenhack]

Many years ago I bought Audio-GD discrete op-amps. I was young and stupid. Did op-amp rolling in a headphone amp and it sounded horrible. Tried two different types, both audible broken. Did measure and realised that they were shit. Later confirmed by Samuel Groner.

I don't understand how companies like Schiit and Audio-gd can continue to push out defective products? Maybe the band Devo was right, we are in a de-evolution.