Hello unpluggged, I think you need to bee more friendly, please do not bite me, I thought this forum is to discuss devices but not arguing with each other.

If I can understand all that manual contain as usual user I think it would be more productive for me, but unfortunately I do not understand some parts of the manual. That's why I shared information in the form that it might be perceived people who are as I'm non-technicians in the first place.

About amplifier, why do I need it for my headphones.

First of all, I believe and I can hear the difference, that each device must provide only one function. For example DAC, preamp, amp, etc.

The second is that you could implement some functions, in RME specifically, which has really good sound, but some which is not. In my opinion the DAC function of RME is a really good sounding element. But if you compare headphone amplifier section of RME to separate headphone amplifiers, you probably discover that separate headphone amplifier sound much better.

The third thing that I want to add tubes to hear warmness of the sound.

In my setup RME balanced headphone output sounds worse than balanced output from McIntosh MHA200 (with Genalec tubes) with Audeze LCD-5. RME headphones output at any volume has distortion which is noticeable (I can hear it) and the sound itself if you compare it with MHA200 has less life in it.

I hope that I answered your question.

Be well.

The sound quality of the ADI does not change with different level reference settings. This is simply not objective "information" of any kind. We don't know what exactly is happening or what you are hearing.

What would be interesting to know is why RME implemented the relais based reference level switching in the first place.

At first sight, it sounds (no pun intended) reasonable as it preserves the SNR but that concept is given up anyway due to the math-based volume control before the DAC - for good reasons.

Now while DACs with analog potentiometers (which aren't common anymore these days due to practical drawbacks such as audible wear; the first Benchmark DAC1's were such a case) indeed preserve the SNR and dynamic range in conjunction to some extent (because they also lower the noise floor when reducing the volume), that theoretical advantage is none in real life as the whole listening situation and one's own ears limit the SNR anyway.

In other words: if not playing at maximum loudness close to the pain threshold, one loses SNR/dynamic range anyway as the upper part partly defines it and based on a constant noise floor, the SNR has to go down as the volume/sound pressure goes down.

The only practical advantage I could imagine might me that in the case of not having those relais switched stages, louder noise sourced might interfere on the way to the power amp and becoming more audible. Clearly, in theory, it is favorable to keep the line level as high as possible for transmission but I still wonder what real relevance that has given today's SNRs of devices.

Don't think so. Since RME's approach audibly and technically most probably already as good as it gets, if I were to really observe any differences, I'd rather suspect the other device intentionally changing the sound or doubt myself (also good from time to time).

Given the pristine technical implementation by RME, there is no sensible explanation why it should sound worse in any way.

Remember: if you can measure a difference, you might (!), often a big might, be able to hear it, but vice versa: if you can't even measure it, it simply isn't there. The claim and mantra that we can hear things which can't be picked up by technical instruments, is audiophile bullshit at its best and should be eradicated. Unfortunately quite the opposite is true: human ears are actually relatively lousy air pressure to electrical/chemical information transmitters and not getting better with age.

Although you could also have that nowadays by the use of DSPs/equalizers, a legit approach if you want essentially what should be called "non high fidelity".