• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of Audio-gd NFB-27.38 DAC and Headphone Amplifier

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,454
Likes
4,274
Location
Seattle Area
#1
This is a review and detailed measurements and comparison of Audio-gd NFB-27.38 DAC and headphone amplifier. It is on kind (and patient) loan from a forum member and ex-Microsoft colleague. The current website price is USD $1,688 plus shipping from China. There is no shopping cart which I assume it means you have to send money via Paypal and then wait for the unit to arrive from China. The loaned unit was purchased in China (?) and is an EU 240 unit powered by a transformer that converts 120 to 240.

The unit is massive by any definition of the word. It is the deepest piece of audio equipment I have seen let alone one for a DAC and headphone amplifier:

Audio-gd NFB-27.38 DAC and Headphone Amplifier Review.jpg


It can swallow four of the Topping DX7s' easily as the one on top of it right now shows. For that humongous size you get every possible input and output. I let you read all the specs for yourself: http://www.audio-gd.com/Pro/Headphoneamp/NFB-27.38/NFB27.38EN.htm. The unit is also very heavy although manageable.

Looking at that page you can see that practically every buzzword for audiophiles is there: ESS 9038 Pro DAC chip, class A operation, tons of power output with discrete amplification, stepped relay volume control, etc. You name it, and it is in there.

The user interface is cryptic due to merging of multiple elements in the same 7-segment display. Changing between DAC and Headphone mode for example is indicated by that HP/PP indicator. Once learned, it is not bad but definitely strange to see "264" which means input 2 and volume level of 64 (out of 99).

There are also tons of jumpers inside for changing configurations such as filter mode and such. These should have been in the UI and not through jumpers. I tested the unit as is without messing with any of that.

For this review I compared the Audio-gd to another Chinese brand, Topping DX7s. Both use ESS DAC chips with the Audio-gd using the more expensive "Pro" variety vs "Q2M" in Topping DX7s. Price wise is no match: the Topping DX7s retails for $499 but goes on sale for $370 shipping included. Still, I thought it made sense to see how two different shops in China produce a product with similar functionality, both aimed at higher end of the performance curve.

I am sure you are anxious to see the measurements so let's get into that. If you are not familiar with my measurements, I suggest reading my tutorial on audio measurements here: https://www.audiosciencereview.com/forum/index.php?threads/understanding-audio-measurements.2351/

For this testing, I was hoping to measure both USB and S/PDIF performance. Alas, I could not get Windows 10 Creator's edition to recognize Audio-gd as a proper sound device. It would recognize it on "hot plug" but then error out in the device manager saying it needs additional installation:

upload_2018-3-23_13-0-9.png


upload_2018-3-23_13-0-40.png


They buy and OEM usb module called "Amanero" and hence that designation above. A reboot and multiple connection/disconnection did not make a difference. There are no drivers on Audio-gd web site. They send you to a web page for that module which I did not go to. Audio-gd says this is UAC Class 2 compliant and works without drivers on Mac and Linux. They need to test this on Windows and fix whatever bug there is.

Measurements
Let's start simple with frequency response measurements starting at maximum volume and reducing it to a few steps and seeing how well the channels match:

audio-gd NFB-27.38 DAC frequency response.png


Only thing to note here is the roll off at maximum volume. This indicates lower than needed gain-bandwidth for the amplifier. Said more simply, they should have kept the max output a bit lower than they let it go.

Channel matching while not perfect, is very good. I think I measured 0.06 dB for the worst-case scenario above. As a reference, my listening tests were performed at the graph second from bottom.

The output level of the unbalanced output, was way, way out there at some 11 volts (RMS). So for the rest of these tests I matched it to the max output of Topping DX7s (around 6 volts). Such was the case in jitter and noise evaluation using J-Test signal at 24-bits/48 kHz:

Audio-gd NFB-27-38 DAC SPDIF Jitter Measurement.png


This is pretty poor performance (and was even worse at higher volumes). We have copious amount of distortion spikes at both < 2 kHz on the left and other ones popping up around 6.5 kHz and 18 kHz. Noise floor is also 10 dB higher. This would be unacceptable performance in a $100 DAC let alone one that costs 17 times as much.

Unfortunately that theme continued throughout the rest of the measurements such as THD+N versus frequency:

Audio-gd NFB-27-38 DAC THD+N Distortion Measurement.png


I could not believe my eyes. This is stunningly poor performance, underperforming the Topping DX7s by some 50 to 60 dB! No doubt this is due to lack of negative feedback. Even hand-picking transistors as they say saves you from poor performance due to non-ideal transformer characteristics of these parts.

We can see the problem in frequency domain testing of noise and harmonics of a 1 kHz tone (itself filtered out):

Audio-gd NFB-27-38 DAC 1 kHz residual distortion Measurement.png


We see copious amounts of both second and third harmonics. They rise up to some -50 dB which again, is worse than Topping DX7s by some 60+ dB. The same higher noise floor is visible as with jitter test. Note again that this is well below maximum output level. The situation gets worse if turned the volume to max.

We would expect problems in intermodulation tests and we see that in this SMPTE signal vs level:

Audio-gd NFB-27-38 DAC SMPTE Intermodulation Measurement.png


Distortion is rising at -20 dB below full scale and before that, noise dominates. The gap again is many tens of dBs relative to Topping DX7s.

How is low level linearity?

Audio-gd NFB-27-38 DAC linearity Measurement.png


I guess not the worst I have measured here but still, the unit loses it once it achieves around at around -90 dB. Not surprisingly, the same problem exists and then some in time domain:

Audio-gd NFB-27-38 DAC -90 dB Measurement.png


That is stunningly bad performance on the left on behalf of Audio-gd NFB-27.38. Not only is it full of noise but occasionally it would go nuts in one or the other channel and produce a taller peak or valley as you see in yellow above.

Let's see if we can make ourselves feel better by looking at output impedance of the headphone jack:

Audio-gd NFB-27.38 DAC output impedance.png


Yeh, we get a bit of good news there with a output impedance a low of 2.5 ohms. Advertised value is 1 ohm. Maybe they use a different frequency than I do but they are close enough to not matter. It certainly beats the Topping DX7s' 10 ohm output impedance.

Let's continue on this track and measure output power versus distortion (THD+N) starting at 300 and 150 ohm loads. Note that these are with both channels driven. They are substantially less than the manufacturer's spec which I assume is taken using one channel and allowance for much higher distortion figures:

audio-gd NFB-27.38 DAC headphone power 300 and 150 ohm.png


I am sorry I got the colors reversed here between the two graphs. In both cases the bottom graph is that of Topping DX7s and the upper one is Audio-gd NFB-27.38. The same massive in distortion amount persists sadly through the headphone output (which I am assuming is the same as what goes out on the RCA jacks in the back). The regular distortion is as much as what Topping DX7s produces while severely clipping!

To its credit the Audio-gd NFB-27.38 does produce more power, albeit at higher distortion level.

The same theme continues at lower impedances of 100 and 50 ohms:

audio-gd NFB-27.38 DAC headphone power 100 and 50 ohm.png


At 100 ohm, output power is 1 watt on NFB-27.38 at threshold of clipping (where the graph sharply shoots up) compared to 0.3 watts for Topping DX7s.

Finally here are the results for 33 ohm:

audio-gd NFB-27.38 DAC headphone power 33 ohm.png


Here at the onset of clipping, Topping produces a very clean output of 0.6 watts at just 0.005% of distortion whereas the Audio-gd NFB27.38 produces 3 watts but at 0.35% distortion.

Listening Tests
Since I could not get the USB interface working without drivers on Audio-gd NFB-27.38, I used a Topping D10 DAC as the USB to S/PDIF bridge. I then ganged up both the Topping DX7s and Topping D10/Audio-gd combo in Roon so that I could play my reference tracks in sync. Levels were matched using a 1 Khz tone well below max level of Topping DX7s (and hence much, much lower than max level of Audio-gd NFB-27.38). I then used an AB switcher to instantly switch between on device versus another.

Testing was performed using Sennheiser HD-650 and HifiMan HE400i.

I could detect no difference in bass or mid-frequencies. In high-frequencies though, there was a subtle difference. They were a bit more emphasized in Audio-gd which gave a slightly more airy and open sound to Audio-gd. But as I focused on them, I realized the high notes are being distorted with the highs having a "sizzle" to them. That exaggerated their presence which could be to some people's liking. For me, as I focused on the same in Topping DX7s, I realized they were very clean and well defined whereas they were "blown" on the Audio-gd NFB-27.38 leading me to prefer the Topping version better.

I should emphasize that the above is very subtle effect. If I did not focus on it, I would find the two DACs to sound the same. Then again, younger people with better high frequency hearing than me will detect the above effect better than me which may lead them to more preference or dislike as the case may be. If you are older than 50, you can safely go with my impressions. :D

Conclusions
Seeing the massive case, and reading the website of Audio-gd, that they have an Audio Precision analyzer and such, set me up for high expectations regarding Audio-gd NFB-27.38 DAC and headphone amplifier. Alas, that hope was dashed with the first measurement where I saw copious amounts of distortion and noise. It was so bad that at first I did not believe it and had to repeat the tests. Unfortunately the same theme continued resulting in some of the worst measurements of any DAC/headphone amplifier I have tested so far.

Subjectively to my ears I could not detect any advantage to this unit. Correlating with measurements, the high frequencies are distorted by a bit and while that can have a euphoric effect of a "more open" sound, I quickly grew tired of it and disliked it.

Needless to say, I cannot in any shape or form recommend the Audio-gd NFB-27.38. In listening to online reviews I hear people fondly call the company "Audio God." Well, I did not see the properness of that attribution in any testing I performed. Hopefully they make other products that perform well. Sadly, this is not one of them.

As always, questions, comments, corrections, etc. are all welcome.
----
If you like this review, please consider donating funds for these types of hardware purchase 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).
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
1,720
Likes
1,068
Location
Monument, CO
#4
I wonder if this unit is defective, bad bias or something...
 

garbulky

Senior Member
Joined
Feb 14, 2018
Messages
347
Likes
47
#5
I' m glad you got a chance to review Audio Gd products. Now the product is rated as producing exponentially higher power than you are measuring. I wonder what the deal is. I hope to get their Audio GD HE-1 preamp someday.
 
Joined
Nov 29, 2017
Messages
24
Likes
15
#6
Very unfortunate to see. A friend of mine measured the NFB27.38 some time back with a QA401 to similar results, but I had been holding out hope that his unit was some sort of one-off defect. Sadly it appears not.

Also Amir, I spy a MiniDSP EARS in that shot. You getting into headphone measurements?
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
3,526
Likes
1,636
#8
Very unfortunate to see. A friend of mine measured the NFB27.38 some time back with a QA401 to similar results, but I had been holding out hope that his unit was some sort of one-off defect. Sadly it appears not.

Also Amir, I spy a MiniDSP EARS in that shot. You getting into headphone measurements?
Okay, so that is a different picture. Others have had similar measurements points to this being normal for the unit.
 

LarsS

Senior Member
Patreon Donor
Joined
Mar 23, 2016
Messages
481
Likes
148
Location
Sweden, Stockholm
#9
As a previous owner (now in my sons possession) of an Audio-Gd Ref 5 DAC I'm not all that surprised of the results from these measurements.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
3,526
Likes
1,636
#10
So do we have an Asian competitor for Schiit?

Seems AtomicBob over at SBAF has gotten some sub-standard results with some other Audio GD products. Not as far off as Amir's, but nowhere close to spec.
 
Last edited:

bunkbail

Member
Patreon Donor
Joined
Feb 27, 2018
Messages
43
Likes
16
#11
I don't know what to believe anymore. Supposedly DS-based DAC should ace measurement tests, but seeing how bad this one goes I lose hope for Audio-gd R2R-based DACs ones, like my own R2R-11.
 
Joined
Feb 18, 2018
Messages
43
Likes
16
#12
When you see inside that massive case with hundreds of components in the signal path, you have the feeling that it's not going to measure right..:)
 

bunkbail

Member
Patreon Donor
Joined
Feb 27, 2018
Messages
43
Likes
16
#14
I just realized that the NFB in its name stands for no feedback. Is it possible to design a DAC without negative feedback but still measures as well as the Topping ones?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,454
Likes
4,274
Location
Seattle Area
#15
Do you think it worth contacting the maker since the listed specs differ so radically from the results you have?

http://www.audio-gd.com/Pro/Headphoneamp/NFB-27.38/NFB27.38EN_Specs.htm
Yeh, I can't make sense out of their specs/measurements.

I built a new fixture with just one channel measuring at full amplitude to see what it outputs from headphone amp. With my 33 ohm load it can only output 3.3 watts or so (about 10.5 volts). They don't have a spec for 33 ohm but they say at 50 ohm where power should be lower, it is supposed to output 8 watts. Maybe in balanced mode power doubles??? Even so, I still can't get their numbers.

I then went after their distortion measurements. They show this harmonic distortion test with similar model Audio Precision to what I have:

upload_2018-3-23_16-22-33.png


First thing I noticed is that the analyzer is in "A-A" mode meaning analog output, analog analysis. Odd because this bypasses the DAC. The unit has analog input so I went ahead and tested it the same way as them, matching every parameter I can see and get this:

upload_2018-3-23_16-23-40.png


As you see, the harmonic distortion I see is in another planet. Their graph says second harmonic at -110 dB, mine is at -52 dB!!!

I tested the same thing for the DAC and got similar results to what I have:

upload_2018-3-23_16-27-54.png


This is all at 0 dBV as they state in the spec:

upload_2018-3-23_16-28-39.png


The only difference is that my load is 33 ohm and theirs is 30 but that should make it easier on them, not hard.

I can't imagine a no-feedback design doing as well as they show. Perhaps the above is a subsystem measurement seeing how it is analog to analog measurement and perhaps the design has evolved???

Hopefully they see these measurements and can respond.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,454
Likes
4,274
Location
Seattle Area
#18
Also Amir, I spy a MiniDSP EARS in that shot. You getting into headphone measurements?
That was the hope when I bought it. But so far, I find it, and all the other measurements out there to be no better than wet thumb in the air to determine wind speed. :) When I can make sense out of something these devices produce, I will start measuring headphones with them. So far I am really disappointed on how primitive the state of headphone measurement is.
 
Joined
Nov 29, 2017
Messages
24
Likes
15
#19
That was the hope when I bought it. But so far, I find it, and all the other measurements out there to be no better than wet thumb in the air to determine wind speed. :) When I can make sense out of something these devices produce, I will start measuring headphones with them. So far I am really disappointed on how primitive the state of headphone measurement is.
Aheheh, well, the EARS itself is pretty intrinsically compromised - in particular in regards to its stock compensation curve, but a case can be made that it has some critical and unavoidable flaws at a physical level - but that's a bit of a harsh take on headphone measurements in general IMO. While there's still a lot of ground to be broken and trails to be blazed, I would argue that the issues of headphone measurements at current are mostly in regards to performance targets, and perhaps to a moderate degree a lack of some supplementary testing protocols (I'd particularly like to see measurements of response variation on a range of human heads, vs. being characterized only on an IEC-compliant HATS), but with the core stuff pretty well locked down.

I'm a member of a small community that's been trying to work out a more refined solution for headphone measurements for DIYers, reviewers, small brands, etc. We've had some discussion and analysis of the EARS - including some attempts to rectify the issues with its stock compensation data. If you'd have any interest, I could send you an invite and/or a copy of what we've got so far.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,454
Likes
4,274
Location
Seattle Area
#20
I'm a member of a small community that's been trying to work out a more refined solution for headphone measurements for DIYers, reviewers, small brands, etc. We've had some discussion and analysis of the EARS - including some attempts to rectify the issues with its stock compensation data. If you'd have any interest, I could send you an invite and/or a copy of what we've got so far.
I would love to see that so yes. :)
 
Top Bottom