Does the AMP/DAC actually influence audio quality
My vote for "diminishing returns" is the $8 Apple USB-C dongle, provided it has enough power. Distortion and noise are too low to be audible under normal listening conditions, performance isn't load-dependent, and output impedance is near zero. It's a great example of how good a cheap chip can be when implemented properly. The $30 Meizu MasterHifi dongle is the next step up, with more power than the Apple dongle and measured performance generally better than the popular $200 stacks (Schiit, JDS, Cavalli) (aside from the comparatively limited maximum power).
With respect to where diminishing returns lie, that's largely down to how much a user is willing to pay to deal with vanishingly audible effects. Keep in mind thresholds of audibility often have a lot to do with how well-trained the listener is, whether the listener knows what cues to listen for, and the type of content used. Auditory masking also plays a huge role: effects that are not audible in music generally may become audible with test tones or specific tracks.
Here are some of the commonly measured aspects of a DAC/Amp that are audible. Roughly ordered in terms of decreasing importance to the average user:
-
Output Impedance. High amplifier output impedances are easily audible with most dynamic headphones since they create a wide bass boost at the driver's fundamental frequency, usually in the "muddy bass" region. I say easily, but that's relative; it's still not incredibly obvious most of the time. Because it's a low-Q resonance, this effect is readily apparent even with standard music.
- Honestly this is a matter of subjective preference. For some users and some headphones this type of boost may be desirable. Spun positively, it can make headphones sound "warmer". Looking at you, tube and vintage amps. IMO it's better to use digital EQ than rely on output impedance.
- Motherboards almost always have a 75Ω output impedance, which means the step up from a motherboard to a basic amplifier is often audible.
- Some headphones do not have this issue. Planars will have significantly reduced power but usually no other audible effects when used with high output impedance amplifiers. The M50X has abnormally little low frequency driver resonance for a dynamic headphone, which makes it sound about the same even when paired with the awful headphone outputs on motherboards and some cheap "pro audio" marketed devices.
-
Channel balance. Some devices will have significant differences between the left and right volumes, usually due to poor potentiometer tolerances. Although this is often measured, it's almost entirely due to manufacturing variations outside of the amplifier maker's control; the measurements are informative with respect to a particular unit but not to the product's performance in general.
- Relay attenuators are a form of analog volume control with no channel imbalance (if designed properly), but are expensive, easy to mess up the design of (in audible ways), and make a clicking noise when the volume is adjusted (which some people like but I don't).
- Stepped passive attenuators are basically potentiometers with a lot of discrete resistors. No channel imbalance, but bulky, expensive, and not as smooth-feeling when turned.
- Digital volume control has no channel imbalance, but can have reduced noise performance compared to a potentiometer unless the DAC is better than the amp. It also can't be practically implemented in an amplifier without other negative side effects, and most users prefer to control their volume from the amplifier.
-
Frequency Response. Not an audible issue on most modern devices. Some niche cases, usually either involving improperly sized coupling capacitors or lossy audio formats, can result in very audible issues – but I'd classify these more as fatal design flaws than legitimate issues.
- I don't count intentional EQ as an inherent issue. Normally devices will be upfront about not being neutral when they use an intentional EQ.
-
Noise. Except at the highest listening levels, most amplifiers are limited by noise (but this noise will usually be below audibility). The regime on the left side of a THD+N vs power graph, where the line is straight and tilted downwards, represents the power levels at which a device's performance is noise-limited.
- On modern devices this is usually constant white noise, from the random fluctuations of electrons in resistors and other resistive elements. This is usually not audible without a lot of gain. If you use significant digital attenuation/EQ and/or listen to very high dynamic range (low average level) content and/or use an extremely sensitive headphone, this may be audible even on decent devices (though this is an atypical use case).
-
Harmonic Distortion. Less audible than spec sheets would like you to believe.
-
Most people won't hear THD below -40dB (1%) with music. It subjectively sounds like "hardness", and is certainly something you can train to hear. Most people can't hear below -60dB (0.1%) with music, even with training.
- It may even sound like increased "detail", a common problem in audio. Poor designs that create audible correlated artifacts sometimes get touted as "more transparent", and this isn't limited to distortion. Looking at you, tube and vintage amps.
- Harmonic distortion is tested at higher output levels than standard listening, and distortion increases with level. In other words, products usually have better practical distortion performance than measurements imply.
- There is an online blind test to check your threshold of audibility. There are some subtle non-distortion "tells" from the way the test's processing was designed, so unfortunately it's not truly a blind test once you figure them out.
- It's a very popular measurement though as it's easy to do and tells a lot about how good the circuit designer is. If I can see the distortion profile (and distortion vs power) of a device and know what chips they used, I can usually make some good guesses about how the circuit was designed (and what mistakes/cost-cutting decisions they made).
- As an amplifier designer I still use THD+N as my design goal most of the time, because it's the most impressive single spec engineering-wise. It may not be audible, but it's still fun.
-
Most people won't hear THD below -40dB (1%) with music. It subjectively sounds like "hardness", and is certainly something you can train to hear. Most people can't hear below -60dB (0.1%) with music, even with training.
With respect to the K5 vs the Schiit Stack: under our current scientific understanding of human hearing, there is unlikely to be any readily apparent audible difference between the two under normal headphone listening conditions. The K5's relatively high noise floor is by far its biggest weakness, so if you can't hear that then there isn't much reason to upgrade.
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now