• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Audio-gd NFB2 192 DAC

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
https://clarityfidelity.blogspot.com/
www.0db.co.kr/xe/
https://m.blog.naver.com/PostList.nhn?blogId=gre_nada
https://www.innerfidelity.com/resources
http://rinchoi.blogspot.com/
https://www.seeko.co.kr
www.nssc.re.kr/phone/19278
http://goldenears.net/board/index.php
Innerfidelity has a bit higher noise floor
goldenears don't have distortion
nssc is very limited
the rest are pretty good

In terms of distortion.
Dynamic over ear type: Sennheiser is pretty good for higher frequencies. Focal is really impressive at lows.
Balanced armature: best is around 0.2% so not that great. 0.5%-1% is also normal.
Planars: mostly around 0.03% probably lower
In ear dynamic: Moondrop kanas, Tanchjim oxygen, Nature sound ns5. these Chinese single dynamic type inears have relatively lowest distortion across the range in Dynamic category. Can go into 0.00x% range at lower volume.
Electrostatic: most around 0.02% or lower
 

Attachments

  • Screenshot_20190616-003831.jpg
    Screenshot_20190616-003831.jpg
    131.6 KB · Views: 227
Last edited:

MetalheadRich

Member
Joined
Dec 28, 2018
Messages
40
Likes
12
I usually very much agree with Amir's sane and reasonable conclusions about his excellent benchmarks, but in this particular circumstance I'd like to play devil's advocate for a moment, and reflect on the meaning of these measurements and on what they might or might not reveal.

Feedback is definitively a very well established circuit design technique, it works well by essentially subtracting the distortion as a difference of the input and the output signal. As I understand it, the criticism to using feedback comes from the fact that there is a significant enough delay in time between the input and the output, and depending on the circuit topology, the signal propagation speed and bandwidth, subtracting the difference signal might not be beneficial under all circumstances.

When measuring the circuit in a steady state situation, and where the test signal is relatively simple, the feedback circuit can virtually remove all distortion. My question is what happens on transient signals? What is the effect of the feedback when the correction does not reflect the current value of the input signal? By how much will it increase the actual distortion? Is it possible to estimate the feedback error or to measure it? Can we came up with a more "dynamic" measurement that can measure distortion under transient conditions?

Finally, it looks like the highest distortion peak stays below 75dB, THD+N is less than two parts part in ten thousand, a number that for a feedback free design is quite remarkable.

Thanks Amir for sparking this very interesting discussion subject, and thanks for having established this awesome site, this place is like is a breath of fresh HIFI air that I have been missing in the last 20 years.
- Fabio

The only steady-state condition is DC. Testing started back in the '70s (TIM) which showed that there are no audible negative effects of NFB. Let's not re-fight old wars.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,550
Location
Seattle Area
Audio gd don't claim good measurements they claim good sound so what's the goal of making a review to say that it measures bad without even trying to see if indeed it sounds good?
They actually do both. Countless people buy their products because they think they also measure well. Why do you think people send them in to be measured?

See this review for example where they even publish Audio Precision measurements which don't seem to show anything close to what I measure: https://www.audiosciencereview.com/...audio-gd-nfb28-28-dac-and-headphone-amp.5147/

index.php


Here is an example for their website now: http://www.audio-gd.com/Pro/Headphoneamp/NFB28.38/NFB28.38EN.htm

1560611968174.png


Until I started to rigorously review their products, most people though they presented the best of both worlds: great measurements and great sound. There were riots in the streets after my first review saying their gear that I have tested doesn't measure well.

Let's say what you say is true. They still need to provide measurements for their devices so that the consumer is well aware of what they are buying. There is "not great" and "horrible" measurements. It is not like the consumer is fully informed with what you say.

As to it sounding better or not, they need to demonstrate that, instead of just claiming it. As we have shown with measurements, what they claim doesn't seem to hold water.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,550
Location
Seattle Area
Feedback is definitively a very well established circuit design technique, it works well by essentially subtracting the distortion as a difference of the input and the output signal. As I understand it, the criticism to using feedback comes from the fact that there is a significant enough delay in time between the input and the output, and depending on the circuit topology, the signal propagation speed and bandwidth, subtracting the difference signal might not be beneficial under all circumstances.
That criticism is not valid. Electricity moves near speed of light. So there is not the kind of "delay" people are thinking about. These are not mechanical systems with slow response.

If there were such delay, then a sine wave would show it as the amplifier distorts differently often depending on what part of the sine wave is going through it. If there was delay and hence lack of correction, our measurements would show it.

Yes you can run out of gain at high bandwidths but that is not a problem here. Plenty of DACs get there with no issues whatsoever.
 

Olli

Senior Member
Forum Donor
Joined
Mar 6, 2018
Messages
331
Likes
242
This is a review and detailed measurements of the Audio-gd NFB2 192 khz DAC. It is on kind loan from a member. It costs US $450 from Audio-gd website.

The overall package while heavy and functional, is not that exciting:

The back panel has the usual connection plus the current mode "ACSS" proprietary connectors which I did not test:


I see no safety or regulatory marks on the unit which is concerning for mains operated units (i.e. NOT using external power supplies).

The unit was plug-and-play on Windows although oddly exposes inputs in addition to outputs! That threw off ASIO4ALL wrapper I use to talk to it in my analyzer. Once I shut down the input, it worked fine.

"NFB" stands for no negative feedback which Audio-gd goes to great lengths to sing its virtues:

View attachment 27681

DAC Audio Measurements
As usual, we start with our dashboard using USB Input:
View attachment 27682

We have about 1 dB of headroom above our nominal output voltage of 2 volts which is nice. Reducing the input by 1 dB did not materially change the performance. High harmonic distortion severely limits performance. This will increase the energy in higher frequencies and will cause some brightness/harshness although most people probably can't tell.

The SINAD (signal over noise and distortion) puts the NFB2 192 squarely in forth (worst) quarter of all DACs tested:

View attachment 27684

Ironically it is not quite as bad as some of their newer designs.

Signal to noise ratio is good but falls way short of specifications:
View attachment 27685

You should be able to play 16 bit signal with the content dominating the noise level, not the DAC.

Linearity by our standards is poor:
View attachment 27687

But again, good enough for CD/16 bit playback.

Jitter shows serious engineering issues but not of audible concern:
View attachment 27688

Intermodulation distortion follows the same story:

View attachment 27689

The DX3 Pro at half the price has far lower noise and distortion.

I am going to stop here and put the energy toward other products to review.

Conclusions
I go into every review/measurement with a fresh slate. Alas, Audio-gd is one company whose measurement outcome can easily be predicted based on previous units tested. Instead of focusing on transparency and excellence in engineering execution, they follow audiophile myths into a ditch. Who says feedback is bad for the ear? Show me the controlled test of this DAC versus one using feedback where the Audio-gd comes out ahead. It doesn't and won't exist.

Fortunately audiophiles are blind to these types of distortions especially as SINAD approaches 80 dB. Hence the reason subjective, non-critical reviews don't hit on the deficiencies in NFB 2 192. Instead of going by sound waves, they substitute their expectation and prejudices in audio and think DACs like this, sound better. Being a false conclusion, that will wear off in later listening, sending them on upgrade path to yet another DAC. Oh well, I will stop the rant here :)

Needless to say, I can NOT recommend the Audio-gd NFB 2 192. There are plenty of other choices at lower costs that run circles around this unit.

----------------------
Questions, comments, critique, etc. are welcome.

In the last review, I mentioned that a work related accident had resulted in decapitation of one of my beloved pink panthers. Some of you wrote in, doubting the veracity of this story. Hopefully the review photo here shows that what I had said was no joke. While some kind donations were made to bring the poor little guy back to health, more money for after-care is needed for nurses and such. So please donate generously using:

Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
I usually very much agree with Amir's sane and reasonable conclusions about his excellent benchmarks, but in this particular circumstance I'd like to play devil's advocate for a moment, and reflect on the meaning of these measurements and on what they might or might not reveal.

Feedback is definitively a very well established circuit design technique, it works well by essentially subtracting the distortion as a difference of the input and the output signal. As I understand it, the criticism to using feedback comes from the fact that there is a significant enough delay in time between the input and the output, and depending on the circuit topology, the signal propagation speed and bandwidth, subtracting the difference signal might not be beneficial under all circumstances.

When measuring the circuit in a steady state situation, and where the test signal is relatively simple, the feedback circuit can virtually remove all distortion. My question is what happens on transient signals? What is the effect of the feedback when the correction does not reflect the current value of the input signal? By how much will it increase the actual distortion? Is it possible to estimate the feedback error or to measure it? Can we came up with a more "dynamic" measurement that can measure distortion under transient conditions?

Finally, it looks like the highest distortion peak stays below 75dB, THD+N is less than two parts part in ten thousand, a number that for a feedback free design is quite remarkable.

Thanks Amir for sparking this very interesting discussion subject, and thanks for having established this awesome site, this place is like is a breath of fresh HIFI air that I have been missing in the last 20 years.
- Fabio
This is a review and detailed measurements of the Audio-gd NFB2 192 khz DAC. It is on kind loan from a member. It costs US $450 from Audio-gd website.

The overall package while heavy and functional, is not that exciting:

The back panel has the usual connection plus the current mode "ACSS" proprietary connectors which I did not test:


I see no safety or regulatory marks on the unit which is concerning for mains operated units (i.e. NOT using external power supplies).

The unit was plug-and-play on Windows although oddly exposes inputs in addition to outputs! That threw off ASIO4ALL wrapper I use to talk to it in my analyzer. Once I shut down the input, it worked fine.

"NFB" stands for no negative feedback which Audio-gd goes to great lengths to sing its virtues:

View attachment 27681

DAC Audio Measurements
As usual, we start with our dashboard using USB Input:
View attachment 27682

We have about 1 dB of headroom above our nominal output voltage of 2 volts which is nice. Reducing the input by 1 dB did not materially change the performance. High harmonic distortion severely limits performance. This will increase the energy in higher frequencies and will cause some brightness/harshness although most people probably can't tell.

The SINAD (signal over noise and distortion) puts the NFB2 192 squarely in forth (worst) quarter of all DACs tested:

View attachment 27684

Ironically it is not quite as bad as some of their newer designs.

Signal to noise ratio is good but falls way short of specifications:
View attachment 27685

You should be able to play 16 bit signal with the content dominating the noise level, not the DAC.

Linearity by our standards is poor:
View attachment 27687

But again, good enough for CD/16 bit playback.

Jitter shows serious engineering issues but not of audible concern:
View attachment 27688

Intermodulation distortion follows the same story:

View attachment 27689

The DX3 Pro at half the price has far lower noise and distortion.

I am going to stop here and put the energy toward other products to review.

Conclusions
I go into every review/measurement with a fresh slate. Alas, Audio-gd is one company whose measurement outcome can easily be predicted based on previous units tested. Instead of focusing on transparency and excellence in engineering execution, they follow audiophile myths into a ditch. Who says feedback is bad for the ear? Show me the controlled test of this DAC versus one using feedback where the Audio-gd comes out ahead. It doesn't and won't exist.

Fortunately audiophiles are blind to these types of distortions especially as SINAD approaches 80 dB. Hence the reason subjective, non-critical reviews don't hit on the deficiencies in NFB 2 192. Instead of going by sound waves, they substitute their expectation and prejudices in audio and think DACs like this, sound better. Being a false conclusion, that will wear off in later listening, sending them on upgrade path to yet another DAC. Oh well, I will stop the rant here :)

Needless to say, I can NOT recommend the Audio-gd NFB 2 192. There are plenty of other choices at lower costs that run circles around this unit.

----------------------
Questions, comments, critique, etc. are welcome.

In the last review, I mentioned that a work related accident had resulted in decapitation of one of my beloved pink panthers. Some of you wrote in, doubting the veracity of this story. Hopefully the review photo here shows that what I had said was no joke. While some kind donations were made to bring the poor little guy back to health, more money for after-care is needed for nurses and such. So please donate generously using:

Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).

the picture with the headless panther is super funny :-D
 

Tup3x

Active Member
Joined
Nov 17, 2018
Messages
162
Likes
130
Location
Finland
Maybe I should give more credit to Realtek... Considering that average mobo the implementation cost is probably closer to 1 €. Judging by that the value is actually mind boggling... Probably even more so on boards where the budget is 2 €. o_O
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
That criticism is not valid. Electricity moves near speed of light. So there is not the kind of "delay" people are thinking about. These are not mechanical systems with slow response.

If there were such delay, then a sine wave would show it as the amplifier distorts differently often depending on what part of the sine wave is going through it. If there was delay and hence lack of correction, our measurements would show it.

Yes you can run out of gain at high bandwidths but that is not a problem here. Plenty of DACs get there with no issues whatsoever.

As movement of electrons is non linear their linear speed is actually app. 1/3 of speed of light, but your argument is of course still valid.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,550
Location
Seattle Area
As movement of electrons is non linear their linear speed is actually app. 1/3 of speed of light, but your argument is of course still valid.
Actually electrons move at snail's pace. It is electricity that moves at that speed, NOT the electrons. The electrons bump into each other and convey the flow but they move very slowly themselves. Sort of like air molecules with respect to sound. This is called drift velocity. There are online calculators for this. For example: http://hyperphysics.phy-astr.gsu.edu/hbase/electric/miccur.html#c1

1560628605607.png


That is just 5 centimeters (2 inches) per hour!!! You would wait quite a while for electrons to move from your amplifier to the speaker terminal!!!

That aside, the velocity factor of electricity relative to speed of light depends on material. Here is the wiki on some RF coax cables:

1560628809451.png


So it ranges from 0.66 to 0.99 ratio to speed of light and hence my comment about "near."
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,288
Location
Gold Coast, Queensland, Australia
That is just 5 centimeters (2 inches) per hour!!! You would wait quite a while for electrons to move from your amplifier to the speaker terminal!!!

The cable guys should jump on that to justify their burn-in times.

After all, you have to get the old factory electrons out of the cable and get new, active electrons in, before cables reach their maximum potential in your system. :)
 
Last edited:

FrantzM

Major Contributor
Forum Donor
Joined
Mar 12, 2016
Messages
4,339
Likes
7,739
There was back in the days a person who spoke about "programming" the electrons to be less random .... or ?? He had a clock that sold for $350 that was basically a $10 Radio Shack clock ... Some people did find this "treated" device to work wonders on their soundstage .. at that time the background wasn't of any shade so no mention of it in the reviews.. ( "dark background", "noise floor" and "Organic" are relatively recent additions to the audiophile vocabulary .... ) Most people didn't find the darn thing to work but some did .... You must read this people.. Please!

https://groups.google.com/forum/#!topic/rec.audio.pro/Mgx5ezUjUGA OMG :eek::rolleyes:o_O:p:facepalm::facepalm::facepalm::D
 
Last edited:

AndrovichIV

Active Member
Forum Donor
Joined
May 25, 2019
Messages
158
Likes
218
There was back in the days a person who spoke about "programming" the make the less random .... electrons to do ... whatever ?? He had a clock that sold for $350 that was a basically a $10 Radio Shack clock ... Some people did find this work on their soundstage .. at that time the background wasn't of any shade so no mention of it in the reviews.. ( "dark background", "noise floor" and "Organic" are relatively recent addition to the audiophile vocabulary .... ) Most people didn't find the darn thing to work but many did .... You must read this people.. Please!

https://groups.google.com/forum/#!topic/rec.audio.pro/Mgx5ezUjUGA OMG :eek::rolleyes:o_O:p:facepalm::facepalm::facepalm::D

This has to be audiophool peak. Nothing will ever amaze me anymore in the world of audio
 

deafenears

Senior Member
Forum Donor
Joined
Feb 2, 2019
Messages
398
Likes
476
Until I started to rigorously review their products, most people though they presented the best of both worlds: great measurements and great sound. There were riots in the streets after my first review saying their gear that I have tested doesn't measure well.
Ah, it all makes sense now... this review is what's behind the recent riots in the streets of Hong Kong ;)
 

NDRQ

Active Member
Joined
Nov 24, 2018
Messages
178
Likes
247
@Duckeenie
@amirm
ASR should be more rigorous and confirm that the audio gd dac don't sound better by validating with a blind test.
If not you re biased like the audiophiles.
And then for us readers who want the truth we don't know who to trust.
Neither scientists neither audiophiles do blind test.
You re all in the same bias and not open minded.

This is some new..how the hell a fact can be biased? :D
Also why the hell we should blind test with our poor human ears, when an Audio Precision gear can make much better measurements.
Human ear is one of the worst thing to test something, it can easily deceive you, even your mental state or mood can affect your hearing..
Its like you saying don't use laser equipoment for measuring a land, just go measure with your footsteps..

Measuring by ear, more or less only works with headphones or soundbars, because they are hard to measure by tools. But dac and amplifier signal is electronic so easy to measure.
 
Last edited:

Hemi-Demon

Senior Member
Joined
Jan 17, 2019
Messages
435
Likes
514
Any chance that amirm's (I assume for testing consistency) insistence on testing all these DACS using ASIO4all drivers and NOT the dedicated manufacturer dedicated drivers as specified, coul have some impact on testing output?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,550
Location
Seattle Area
Any chance that amirm's (I assume for testing consistency) insistence on testing all these DACS using ASIO4all drivers and NOT the dedicated manufacturer dedicated drivers as specified, coul have some impact on testing output?
No. When there is an issue with ASIO4ALL I either can't use the device, or it truncates to 16 bits. In the latter case, SINAD gets limited to 96 dB. When SINAD is lower, it is always the device that is doing that. And we can see that linearity is pretty good indicating that there is no truncation to 96 16 bits.

I can always verify by playing the test tone through WASAPI but really, it is not necessary in this case. When it is, I double check things that way.
 

Tks

Major Contributor
Joined
Apr 1, 2019
Messages
3,221
Likes
5,494
What’s up with these companies that don’t want to use feedback designs? Another device that was tested said they use ZERO feedback. Bombed in similar fashion to this device. :facepalm:
 

Zog

Active Member
Joined
Mar 13, 2019
Messages
255
Likes
290
In the last review, I mentioned that a work related accident had resulted in decapitation of one of my beloved pink panthers. Some of you wrote in, doubting the veracity of this story. Hopefully the review photo here shows that what I had said was no joke. While some kind donations were made to bring the poor little guy back to health, more money for after-care is needed for nurses and such. So please donate generously using:

A moments silence please.

Pinky was a good panther. For years he served his master and the audiophile community.
A faithful servant, always on hand, never complaining, and very photogenic.
Always ready to give helpful advice, yet never sneering at the users of gold plated power cables.
He would sometimes affect a casual air, but that only endeared him all the more to his beloved community.
It is with sad hearts that we learned of his demise. Struck down in the prime of life.
To his master: we share your grief. We offer our deepest condolences.
Pinky helped us all make a good buy, but now alas it is good bye.
RIP
 

BillG

Major Contributor
Joined
Sep 12, 2018
Messages
1,699
Likes
2,266
Location
Auckland, New Zealand
Is there a single device from China, which is good for something?

Oh, please! Stop with the Sino hate, it's ignorant! There are numerous, well engineered Chinese products on the market, and probably even more from the big names in audio that are manufactured there with input from Chinese engineers... :facepalm:
 
Top Bottom