Do You Need an External DAC/Headphone Amplifier?

[amirm]

Meaning, if the transducer has >0.1% THD across the board, will anyone notice 0.005 vs 0.0005% coming from the electronics?

If your drinking water has too much lead, adding arsenic to it makes a difference or not?

Single value numbers like this cannot be used to make conclusions. This is why I show the full spectrum of these distortions. To the extent you have that for your speaker, then you can analyze them together.

 

[Xyrium]

I wasn’t pleased with it as a headphone amp. I Used it mainly as a DAC

That's exactly how I'm using it. My desktop rig is: STX II>Nuprime STA-9>Dynaudio Emit 10s. Sound is excellent. I'm missing the bottom octave, but overall clarity, especially in the midrange, along with good snap in the lows, works well for such a low-budget system. For some reason, the preamp section of the STX seems to work very well with my STA-9.

This said, I have my doubts that EMI is impacting performance here. How would one look for the effects if it were present in the audio stream? Would it be a buzzing sound, humming sound, measurable distortion across the spectra?

 

[Sythrix]

This said, I have my doubts that EMI is impacting performance here.

Impacting performance isn't quite the way I would word it. There is going to be EMI and other noises inside of a system, dependent on the components inside and their proximity to the card, etc. That said, Asus is not stupid. They have thought of this and have factored it into their design. If we had multiple systems with various components and configurations, we might be able to measure it, but I wouldn't worry about the impact being audible. I have never been able to hear buzzing or anything audible across both of my builds that featured the card and with many different components and configurations.

Also, I have almost exclusively used the headphone output (except for a brief spell where I had my DIY Butte amp hooked up) and have no complaints whatsoever. However, I've always had 250 - 600 ohm headphones on my computer, so the output impedance hasn't affected anything. I would be interested if it actually measured at ~10 ohms once Amir gets his hands on it.

 

[Dana reed]

If your drinking water has too much lead, adding arsenic to it makes a difference or not?

Single value numbers like this cannot be used to make conclusions. This is why I show the full spectrum of these distortions. To the extent you have that for your speaker, then you can analyze them together.

Well distortion doesn’t have toxicological properties that can make you sick. A better analogy would be tasting a little bit of mayo added to your blue cheese dressing. Probably not easy to do.

 

[Xyrium]

Impacting performance isn't quite the way I would word it. There is going to be EMI and other noises inside of a system, dependent on the components inside and their proximity to the card, etc. That said, Asus is not stupid. They have thought of this and have factored it into their design. If we had multiple systems with various components and configurations, we might be able to measure it, but I wouldn't worry about the impact being audible. I have never been able to hear buzzing or anything audible across both of my builds that featured the card and with many different components and configurations.

Also, I have almost exclusively used the headphone output (except for a brief spell where I had my DIY Butte amp hooked up) and have no complaints whatsoever. However, I've always had 250 - 600 ohm headphones on my computer, so the output impedance hasn't affected anything. I would be interested if it actually measured at ~10 ohms once Amir gets his hands on it.

Indeed, I have never heard the effects of EMI, if it did exist in any of my builds. I usually use premium parts (PSU, wiring routes, etc) when possible, and MB manufacturers have become even more cognizant of this claimed issue in recent years. Let's remember, before USB and Firewire, most studios were using onboard cards with little to no impact from the claimed electronic noise inside of a PC case. Even low end cards like the venerable EMU 1212M have been used by many with zero impact on measurements or perceived sound quality.

I am surprised that Asus missed the mark on the output impedance of that card, it seems to do well otherwise! If Amir has his hands on one, I'd be very curious as well!

 

[Sythrix]

Indeed, I have never heard the effects of EMI

If you want to, run the stock front-end headphone cable across the system with no routing (like most people would). You'll hear it as a buzzing, typically. The more components you have, the worse it is. Annoying as all hell. I never run that cable anymore after I found out how terrible they were. I sincerely doubt any of them are even shielded, or very poorly if at all.

It wouldn't shock me if this was how most newbie system builders were interpreting their sound situation.

EDIT: I'm referring to the cable that says "HD Audio" or something on it. Also connects a front-end MIC. Poor spindly little thing.

 

[Blumlein 88]

I've measured internal sound cards that have a connection to front panel jacks on desktop PCs. Those front panel jacks usually have more noise. Sometimes it is ultrasonic, but most have 10-20db more noise up there vs using the jacks at the back of the card. So I presume it is picking it up inside the case from running around/through the loom for all the other activities in the machine.

 

[Sythrix]

So I presume it is picking it up inside the case from running around/through the loom for all the other activities in the machine.

That is exactly what happens. The front panel jacks are useless for anyone who even remotely cares about sound quality, unless you are willing to install a new cable with a proper shield (never attempted it, as running from the back has not been a problem for me).

It has been clearly audible; a buzzing in all of the cases where I've tried to use the front panel.

 

[bennetng]

@amirm
https://www.gigabyte.com/Motherboard/GA-Gaming-B8-rev-10#kf
The Gigabyte board under test also has a dedicated USB port for DAC, could you test it as well?

 

[Sythrix]

@amirm
https://www.gigabyte.com/Motherboard/GA-Gaming-B8-rev-10#kf
The Gigabyte board under test also has a dedicated USB port for DAC, could you test it as well?
View attachment 14917

These images are hilarious. I love the featureless, vague pseudo-graphs. Like they squished the power down and now it’s behaving in the middle there.

I would also like to know the results of measuring an (exclusively) usb powered DAC off of this source.

 

[Zilfallion]

These images are hilarious. I love the featureless, vague pseudo-graphs. Like they squished the power down and now it’s behaving in the middle there.

I would also like to know the results of measuring an (exclusively) usb powered DAC off of this source.

And maybe compared against some of the USB decrapifiers.

I do use that port on my motherboard for my Element. Because it's there. At the very least, it doesn't have horrible noise. Or, well, audible noise in my setup.

It DOES have a potential use though, you can disable USB-power to the port in BIOS if you have a self-powered DAC. And for some reason you can also offset the voltage by +0.1, +0.2, or +0.3V, though I never touched that.

 

[Sythrix]

And for some reason you can also offset the voltage by +0.1, +0.2, or +0.3V, though I never touched that.

Probably to combat voltage droop across long or inefficient cables. I can see that being a priority because it may influence the performance of the DAC if it can't pull sufficient voltage.

 

[έχω δίκιο]

The big question, which has been posited here several times but not answered clearly, is: What measurements and metrics will actually be discerned by a listening human (on a given set of transducers and source material).

The audio industry has been struggling to determine that for decades. I think they all matter, but there are going to be thresholds of distortion below which the human ear can't detect.

Meaning, if the transducer has >0.1% THD across the board, will anyone notice 0.005 vs 0.0005% coming from the electronics?

Probably not. But THD lumps all harmonics together, whether they are even, odd, low-order, or high-order harmonics. Two amps with 0.05% THD could have radically different harmonic peak levels if one had many lower level harmonics and the other had just one or two very prominent ones. By analogy, Ivory Soap claims to be 99.44% pure, which is great if the 0.56% impurity is something like vegetable oil, but not so great if it's the Russian nerve agent Novichok.

And if the noise floor of a recording is up at -80 dB, who cares what linearity and sine waves down at -90 or -100 dB look like?

You just reminded me that I wanted to beat someone for ripping vinyl LPs at 24bit/192k resolution.

 

[ylkamalesh1]

This might be a tangent.
I bought a Topping DX7s. I run it with a pair of JBL LSR305, LSR310s & HD6XX.
From what I observed, DX7s definitely makes a difference but, it is very subtle from the PC with ALC889 I was using when I play with TIDAL master source.

I think, the external hardware makes real sense once you start bumping up the source quality beyond 24bit/96khz.
Most consumer hardware is surprisingly good till 24bit/96khz

 

[andreasmaaan]

This might be a tangent.
I bought a Topping DX7s. I run it with a pair of JBL LSR305, LSR310s & HD6XX.
From what I observed, DX7s definitely makes a difference but, it is very subtle from the PC with ALC889 I was using when I play with TIDAL master source.

I think, the external hardware makes real sense once you start bumping up the source quality beyond 24bit/96khz.
Most consumer hardware is surprisingly good till 24bit/96khz

Unless something is wrong, there can't be an audible difference between 24/96 and any higher sample/bit rate.

The ALC889 claims a THD spec of -90dB @ -3dBfs. If that's correct, it might be just audible, depending on the type of distortion (which is unknown) and the system in front of it. The DX7s should not produce any audible distortion under any conditions.

If you can hear a subtle difference between your PC soundcard and the DX7s, this could be the reason.

Another possibility is of course expectation bias

 

[n2it]

@amirm
https://www.gigabyte.com/Motherboard/GA-Gaming-B8-rev-10#kf
The Gigabyte board under test also has a dedicated USB port for DAC, could you test it as well?
View attachment 14917

@amirm, I am curious too if turning the power off to this port makes a difference (and while you are at it, if there is any difference when using the bus power between this port and a regular usb port).

 

[ylkamalesh1]

Unless something is wrong, there can't be an audible difference between 24/96 and any higher sample/bit rate.

I agree. We can achieve for better specs from engineering perspective but for an untrained ear, the differences are super subtle.

 

[έχω δίκιο]

The front panel jacks are useless for anyone who even remotely cares about sound quality, unless you are willing to install a new cable with a proper shield (never attempted it, as running from the back has not been a problem for me).

EMI inside computer cases varies tremendously based on everything from the chosen components to the physical arrangement of the boards, power supply, fans, and cabling.

I assume that fantastic measurements on sound cards, if actually taken, are likely made on computers which bear little resemblance to normal home PCs. I would not be at all surprised to find out that they used tricks like relocating the power supply to an external enclosure and adding braided shields to all power and data cables. They could even be putting the sound card into an external PCIe expansion box, all by its lonesome, possibly with a disconnected fan and with EMI shielding tape at all seams.

 

[andreasmaaan]

I agree. We can achieve for better specs from engineering perspective but for an untrained ear, the differences are super subtle.

I would say non-existent rather than subtle.

In the case of bitrate, the analogue electronics will not be capable of 24 bits so any extra bits the DAC offers won't be seen. And in the case of sample rate, the ear won't be capable of hearing the higher frequencies, even if they are recorded in the first place.

 

[έχω δίκιο]

We can achieve for better specs from engineering perspective but for an untrained ear, the differences are super subtle.

No amount of training is going to let your ear hear frequencies above 48kHz, which is the upper limit of 96kHz sampling rate (see Nyquist's theorem). Nor will you be able to hear bit resolution in excess of 24 bits due to limitations in analog electronics (as andreasmaaan points out), some of which are based on physics, like Johnson–Nyquist noise (AKA thermal noise).