[KraftD1]

I was looking around the portable source gear section and people are talking about this or that DAP sounding better/more detailed/etc.  Also, a lot of advertising/PR seems to talk about how iDevices are crap so you need gadget or DAP 'X'.  However, when I look at measurements over at GE, I get kind of confused. 
 
For example if you take the AK120 and an iPod Touch 5G.  With no load the AK120 has better THD+N, Dynamic Range, etc.  However, if you look at them under load (near the bottom) the iPod has lower THD across the tests (R16 to R250 ohms) and better IMD+N in all tests except R250.  So, if we actually assume those levels are distortion are audible, doesn't that look like the iPod outperforms the AK120 on performance?
 
AK120: http://en.goldenears.net/index.php?mid=GR_Mobile&document_srl=18873
Touch: http://en.goldenears.net/index.php?mid=GR_Mobile&document_srl=16149

 

[mikeaj]

In general, audibility of distortion depends on the kind of distortion products that are generated and the program material, so one number may not necessarily actually represent a better result than a higher one. Also, this is a 1 kHz test; distortion may be different at say a lower or higher frequency.
 
In addition, the test may not be at an equivalent volume level or otherwise a fair direct comparison.
 
That said, the spectrum is very unlikely to be something like almost all easier-to-hear 7th order distortion as opposed to everything else. It's usually dominated by 2nd or 3rd, with 2nd in particular being tough to hear. However, a reading of 0.020% is most likely not something actually audible. Also consider that most headphones are dealing with an order of magnitude worse distortion if not worse.
 
It is unlisted what the absolute noise levels are but just the value referenced to the signal level... that said, the AK120 may offer a lower noise floor which could be potentially useful when using very sensitive IEMs where you need to crank the signal level down quite a bit. It can also actually play 24/96 material natively, for what that's worth (generally nothing audible, but still...)
 
 
Another point to make here is that no, the "crappy i____" does not in fact have poor audio performance. Of course the marketing types want to rail on these things to validate the existence of their products. I might say the most meaningful difference here in terms of audio performance (never mind convenience factor and so on) would be the difference in output impedance, which is relevant for many popular BA IEMs that people are actually using on the go. The 3.5 ohms for the AK120 could change the balance of those a couple dB or so, definitely more so than the 0.75 ohms on the iPod touch 5g.
 
That said, most of the people making arguments for this kind of stuff have low confidence in the relevance for usual audio benchmarks in terms of characterizing sound quality. They're not pointing to the benchmarks or to rigorous listening tests as supporting evidence to their claims of superior quality. Actually, many companies avoid making claims of superior quality directly but make what are more like insinuations and/or just quote users.
 
As for a takeaway? You could say there could be some great unknown forces acting upon audio performance. Or rather, at least something not showing up in the above measurements, which are in fact not at all that comprehensive. Or you could say that avoiding the facts makes for a more convenient truth, especially when you have audio equipment to sell.

 

[KraftD1]

Thanks for breaking it down, appreciate it.

 

[castleofargh]

good specs with no load means that it will sound great when you don't plug anything in it ^_^.

 

[mikeaj]

Well, they're loading it with whatever they're using to record the inputs for RMAA, in a sense...
 
So a line input. Like the performance you might see when using the device to power another amp.

 

[castleofargh]

well I really never asked myself how it was done for that measurement ^_^. I guess I always took it for granted while not knowing anything. I imagined that on the contrary they would use a massive resistivity so the system could be considered as unplugged.
 
anyone who actually knows how it's done can step in plz?

 

[dvw]

A no load test is never really no load. I don't know how exactly they do the test. But in my previous experience we usually use a high impedance to emulate a low load. The idea is to test the amplifier's capability with no load. A load (or different loading) is to test the driver's driving capability. For example if you have a FET driver, the FET could go into current limiting at higher power. So in general, you will see the amplifier has higher THD when it's near its power limit.The other factor that might influence this is the thermal characteristic of the assembled device.
 
It is difficult to form an explicit opinion without knowing the exact design in this case though. Specification can be made up, that is why the specification with test condition is critical for true understanding the devices But I do believe the ipod is competitive in terms of quality from this simple spec.
 
P.S. In ATE testing, a lot of time a current sink or source is used to emulate load.

 

[xnor]

"No load" means usually that the device under test just has to drive the input of the sound card or audio interface used to record the test signal. That load is usually 10 kOhm or higher.
 
In the last tables you can see columns called "R250" ... "R16". That means that a 250 ... 16 ohm resistor was connected parallel to the input of the sound card.