I have had my A17 doubled amped with a Fiio E12A and the A17 volume maxed out for months now and no distortion at all. In fact I find I get the best sound with the A17 maxed vs a lower volume setting when double amping. I believe Sony is using fixed volume output at max volume for use with an external amp as it does not have a true LO implementation. Are you sure you are using your analyzer properly? I think I will trust my ears thank you very much.
I guess it is never too old to use the 'you probably don't know what you are doing' argument
. No offence taken, but let's look at some graphs.
First, let's look at what a typical 1kHz pure tone coming out of a decent DAP looks like. Here is FiiO X5II's headphone-out at volume 88, giving roughly the same voltage as A10's LO. Since it is a 1kHz pure tone, we shouldn't see any extra peaks beside the 1kHz (*and we didn't). Background noise is mostly under -120dBV, which is about -108dB from the 1kHz peak.
Here is A10 are volume 14/30. As I have commented before, this is where A10 behaves at its best. Pushing the volume up eventually will increase distortion.
This is how A10 LO looks like with all EQ and HSEE disable (same condition for all the following graphs) - notice there are a lot of unwanted small peaks after the 1kHz tone. We know these are harmonic distortion because they are mainly the multiplied frequencies (2kHz, 3kHz, 4kHz, etc) of our 1kHz fundamental frequency.
Here is what A10 looks like at 24/30 volume - notice the voltage output is almost the same as LO. Still have lots of harmonic distortion, but just a tiny bit cleaner than LO. Ironic actually.
This is how A10 looks like at 30/30 (maxed out) - rather horrible. You don't need to be an engineer to know why maxing out the volume is such a bad idea. Also take note that 30/30's voltage is in fact almost double that of LO.
Here is A10's LO feeding E12A (with a 47ohm load, set to roughly gain of 1 *no amplification*). E12A is doing such a great job and being so transparent that it reproduces the exact harmonic distortions from A10's LO, hence why this graph looks so alike to the LO graph above.
A10 at 24/30 feeding E12A, looking just like A10 at 24/30 without E12A. Again, E12A is doing its job as described.
A10 at 30/30 feeding E12A, same thing.
Just for good measure, this is A10's LO with HSEE enable, just to show you how much crap HSEE throws into the mix when you compared it to pure LO. Perhaps the only comforting news is that they are mainly even harmonic, which means they are not quite as offensive to the ear.