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

[Killingbeans]

How do I think an FPGA can affect sound quality? I think how it's programmed will likely have an impact on sound quality, I also think how power is handled by the System as a whole will also affect how the FPGA operates and ultimately how it can affect the sound quality.

An FPGA is basically just a huge array of Nand-gates that can be connected to each other in order to do all kinds of boolean algebra. It operates purely in the digital domain. The only way the power supplied to the chip could have any impact on the outputs, would be if it was so catastrophically bad, that it made '1' become '0' and vice versa. It's very unlikely to happen.. unless it has been implemented by a nincompoop

I don't know anything about how Chord products are designed, but I would imagine that the FPGA has a few external transistors to handle the "analog" part of the DA conversion(?). Those would be sensitive to noise. But so would the analog part of any other DAC.

 

[VintageFlanker]

Amir's measurement environment is not carefully crafted btw. He just uses his pretty standard laptop, cheap cables, standard wall socket, no power conditioning, etc.

Yep. And that would be very interesting if @amirm could measures how these factors may or not impact measurements. Analog, digital cables, main blocks etc.

 

[andreasmaaan]

Yep. And that would be very interesting if @amirm could measures how these factors may or not impact measurements. Analog, digital cables, main blocks etc.

Do USB Audio Cables Make A Difference?

The short version: only in extreme cases, and not enough to be audible.

 

[Killingbeans]

Can I ask,

Why a high quality ceramic or any SMD ceramic would be a better choice for audio? Not exactly a fan of ceramic caps unless it’s for RF.

I was pointing at film caps being used as supply decoupling on digital chips, which is silly, since it actually is an RF application. C0G/NP0 SMD caps would be a far, far better choice. Although the most important aspect of using SMD caps would be to make the decoupled trace as short as possible.

I wouldn't recommend any particular type of capacitor to put in an analog audio signal path. It all depends on how it oscillates with the components around it. Depending on how it reacts to the resistance, inductance and capacitance parallel to it (external or parasitic), it will present a more or less desirable impedance curve.

 

[Krunok]

. It all depends on how it oscillates with the components around it. Depending on how it reacts to the resistance, inductance and capacitance parallel to it (external or parasitic), it will present a more or less desirable impedance curve.

 

[Thomas savage]

i find changing cables periodically makes me hear new things.

 

[Krunok]

i find changing cables periodically makes me hear new things.

But not as much as changing avatars, right?

 

[Thomas savage]

But not as much as changing avatars, right?

I change avatars more often than I change my underpants,, true story ...

 

[Krunok]

Ok, I got what I was asking for.

 

[Jimster480]

Fair enough, you're being rational and realistic about all this and maybe not testing things in a way that's scientifically rigorous but doing a few things you can do without turning your home into a test lab

The issue for me still remains this one though: how can a person possibly hear anything 100dB below the signal?

Oh I'm not sure what I'm hearing would be something 100db below the signal. I can't tell you since I'm not measuring it, I can just tell you what I hear by going back and forth.
I think that the analyzer tests don't test every possible combination. For example mfg usually post specs at a certain frequency, like 115db Snr @ 1khz. If amir tests at a different frequency, the performance isn't necessarily the same.


Well with music there are many different tones of all different frequencies, and that must have some effect on the performance of the product.
I think this is where amir's linearity test has more weight than some other ones, as it tests performance across all frequencies.
We were also discussing the reduced performance of amplifiers while driving actual headphones due to feedback.
I think that this might also be the cause of what I have been able to hear, simply because different amp topology would respond differently to the loads of different headphones feedback.

It's part of how Amir test of the liquid carbon was interesting to me, not because I am a fan of that amp or expected it to be better. But when we tested it at my house (me and my couple friends into audio) it was unmistakably warmer sounding than every other amp we had here for testing. But in amir's test it had a flat frequency response... So there has to be something else at play here.

before I ever found this site (and I was stuck on headfi) I had drawn various conclusions through my own testing that some DACs were more resolving than others (atleast in my testing). Later on after finding this site and sending amir a box of audio devices, I have to say that I was spot on with my conclusions vs the test results here. I think that possibly amir should try to develop a similar linearity test for amps, as it could give more insight vs what has been currently tested.

The thing that made me look for this site in fact was due to me purchasing Schiit products (after all with the hype on headfi, you end up trying them sooner or later) and finding that they sounded really bad. In fact my fulla2 had a noticeable drop / pop / crack which occurred during quiet parts of classical tracks, my modi multibit had the same issues... Schiit threatened me when I notified them of what I found. So I originally planned to rent an analyzer and invite all my friends with Schiit gear. Upon making a thread about said meet up, Jude sent friends of his own to derail the thread until it was administratively closed.
I was so angry I actually reached out to quite a few audio brands that I thought were honest and asked them to work with me; they were interested in results but largely "afraid" of the Schiit - headfi partnership exileing them too and ruining their businesses.
Thankfully I found Amir here and he was responsive to what I had to say, and well now the rest is history!
I am so thankful for the great work Amir does here to keep the audio scene in check.

 

[andreasmaaan]

Oh I'm not sure what I'm hearing would be something 100db below the signal. I can't tell you since I'm not measuring it, I can just tell you what I hear by going back and forth.
I think that the analyzer tests don't test every possible combination. For example mfg usually post specs at a certain frequency, like 115db Snr @ 1khz. If amir tests at a different frequency, the performance isn't necessarily the same.


Well with music there are many different tones of all different frequencies, and that must have some effect on the performance of the product.
I think this is where amir's linearity test has more weight than some other ones, as it tests performance across all frequencies.
We were also discussing the reduced performance of amplifiers while driving actual headphones due to feedback.
I think that this might also be the cause of what I have been able to hear, simply because different amp topology would respond differently to the loads of different headphones feedback.

It's part of how Amir test of the liquid carbon was interesting to me, not because I am a fan of that amp or expected it to be better. But when we tested it at my house (me and my couple friends into audio) it was unmistakably warmer sounding than every other amp we had here for testing. But in amir's test it had a flat frequency response... So there has to be something else at play here.

before I ever found this site (and I was stuck on headfi) I had drawn various conclusions through my own testing that some DACs were more resolving than others (atleast in my testing). Later on after finding this site and sending amir a box of audio devices, I have to say that I was spot on with my conclusions vs the test results here. I think that possibly amir should try to develop a similar linearity test for amps, as it could give more insight vs what has been currently tested.

The thing that made me look for this site in fact was due to me purchasing Schiit products (after all with the hype on headfi, you end up trying them sooner or later) and finding that they sounded really bad. In fact my fulla2 had a noticeable drop / pop / crack which occurred during quiet parts of classical tracks, my modi multibit had the same issues... Schiit threatened me when I notified them of what I found. So I originally planned to rent an analyzer and invite all my friends with Schiit gear. Upon making a thread about said meet up, Jude sent friends of his own to derail the thread until it was administratively closed.
I was so angry I actually reached out to quite a few audio brands that I thought were honest and asked them to work with me; they were interested in results but largely "afraid" of the Schiit - headfi partnership exileing them too and ruining their businesses.
Thankfully I found Amir here and he was responsive to what I had to say, and well now the rest is history!
I am so thankful for the great work Amir does here to keep the audio scene in check.

I agree that for many devices Amir (understandably) doesn’t conduct enough tests to draw conclusions about performance in all possible cases, fair point.

Here though we’re talking about the ADI-2 DAC, which Amir measured very comprehensively, and the Mojo, which was measured quite thoroughly by Amir and more comprehensively elsewhere. In these two cases I feel I have enough data to be justified in saying there will be no audible difference in any normal use case.

Both units comfortably stay under -100dB for noise and distortion at all frequencies in the audio band and are very linear down to 19-20 bits.

 

[Krunok]

.. and are very linear down to 19-20 bits.

Whatever that measurement exactly means.

 

[andreasmaaan]

Well with music there are many different tones of all different frequencies, and that must have some effect on the performance of the product.

Last separate little point: this is what the IMD tear that Amir does is for. This test uses tones at opposite ends of the frequency spectrum and gives a very good indication of performance with complex signals (ie music).

You’ll find discussion here about the merits of this test vs other multitone tests; the general consensus is that this test is adequate for gauging a device’s performance with music.

 

[Krunok]

the general consensus is that this test is adequate for gauging a device’s performance with music.

I would say Amirs test go even beyond gauging device's performance with music - they gauge how well the products are engineered.

 

[Jimster480]

I agree that for many devices Amir (understandably) doesn’t conduct enough tests to draw conclusions about performance in all possible cases, fair point.

Here though we’re talking about the ADI-2 DAC, which Amir measured very comprehensively, and the Mojo, which was measured quite thoroughly by Amir and more comprehensively elsewhere. In these two cases I feel I have enough data to be justified in saying there will be no audible difference in any normal use case.

Both units comfortably stay under -100dB for noise and distortion at all frequencies in the audio band and are very linear down to 19-20 bits.

I wasn't discussing either product, just products in general. I've never listened to a Hugo or an ADI-2.

Last separate little point: this is what the IMD tear that Amir does is for. This test uses tones at opposite ends of the frequency spectrum and gives a very good indication of performance with complex signals (ie music).

You’ll find discussion here about the merits of this test vs other multitone tests; the general consensus is that this test is adequate for gauging a device’s performance with music.

I understand that IMD is supposed to do this and I have been in the discussions about multitone tests. But as has been shown before, performance can be different when different tones are being tested.
That's all I was saying.

 

[mshenay]

An FPGA is basically just a huge array of Nand-gates that can be connected to each other in order to do all kinds of boolean algebra. It operates purely in the digital domain. The only way the power supplied to the chip could have any impact on the outputs, would be if it was so catastrophically bad, that it made '1' become '0' and vice versa. It's very unlikely to happen.. unless it has been implemented by a nincompoop

I don't know anything about how Chord products are designed, but I would imagine that the FPGA has a few external transistors to handle the "analog" part of the DA conversion(?). Those would be sensitive to noise. But so would the analog part of any other DAC.

I hear ya! And ones as zeros are defined by a max & min set voltage with a grey area in between right

and to be honest the Hugo 2 review unit that I had for a couple of weeks, as hot as it ran I think catastrophic may have been in his future lol

on the subjects of cables measured making a difference I'd be curious about that too

I will admit that switching away from USB and swapping into a new set of blue jean cables an Etir & all new power cables did make my audio GD a little blacker... the distance a I noticed less noise during silence or even quite passages

 

[Addicted to music]

I would say Amirs test go even beyond gauging device's performance with music - they gauge how well the products are engineered.

But ultimately you won’t know until Amir does a DBT.

When I got back into audio on another forum, boy, the Audio DG fan club was huge. You couldn’t post an investigative comment on the product. Even other forums were the same. One inquisitive post and you have them all over you like vultures picking on you like you’re the bad guy. I knew that this brand didn’t performed with measurements. One post I made I personally had to stay away for a week! This guy who was an absolute dedicated fan turned Mr Hyde and I never reply to any of his posts. The product was when audio GD 1st brought out there Precision Amplifiers. The pics on his website shows 4 ouput devices capable of 15A each and all are mounted on a plate with no heatsink! So it doesn’t have to measure well for proper engineering! Infact you can still see this on his website. Maybe he designed this range of product just to drive horns! But boy I never thought I get that reaction on the WWW.
Everyone that own the product have similar comments about how it sounds; “ more meat on the bone, full of rich harmonics “. Etc etc.... ever heard those comments? If not jump on Johnny Darko’s site and have a giggle! So I just wonder that whether measuring well is an audiophile criteria?

 

[Veri]

“ more meat on the bone, full of rich harmonics “.

they're not wrong ... "Massive 3rd Harmonic Distortion", lol

 

[Krunok]

But ultimately you won’t know until Amir does a DBT.

I don't think so. If DAC measures well on his instrument you don't need a DBT to know it won't affect sound as you hear it.

What I wanted to say is that Amirs criteria for DACs exceeds criteria of non-hearable levels.

 

[Killingbeans]

I'd like to change "parallel to it" to "parallel to it and in series with it" (was half asleep when I wrote it), but what else is nonsense?

It's also pretty simplified to talk about resistance, inductance and capacitance, when ESR and reactance (inductive, capacitive and magnetic) technically are more relevant, and mostly only at RF frequencies, but I'd quickly get in over my head if I choose that route. My point is just that any type of capacitor will do fine in an audio application as long as a set of parameters, that are well known to science, are within spec. There is no type of capacitor that "sound" better than others, and there's certainly no dielectric material like oils or beeswax or some other BS that has these magical properties.

I do hate electrolytic capacitors with a vengeance though. But only because they often have a pathetic lifespan and a tendency to die kamikaze-style.