So tell me this ....

It CAN do that for several reasons actually.



An amplifier can 'add' information in several ways actually.
1: By adding extra harmonics (what non-feedback and all tube amps can do as well as some other designs that have a measurably high amount of distortion.
Not all sorts of added 'distortion' is perceived as distortion (as in sound degrading) but can 'enrich' as well.
Somewhat similar as a vinyl cut can 'add' because of the needed processing before sound can be 'managed' by the cutter head which seemingly improves the sound.
2: By loading an 'amplifier' from a source with a higher impedance load (the input of the amp) which can improve (not by adding but by NOT removing) SQ.
For instance if the source has a low value output capacitance (as many DAP's including ipods do) which creates a high-pass filter in combination with low impedance headphones but does NOT when loaded with the (VERY high) input impedance of an amp so NO-roll-off in the lows which makes it fuller bodied.
Another possible snag is limits of the amp section of the source are reached which increases distortion when loaded with a low impedance load.
When that load is dramatically decreased (by the high input resistance of an amp) the distortion can decrease dramatically and drop under 'detectable levels'.
That would increase the SQ but isn't caused by adding info but by not removing it.



In many blind tests removing information from the audio signal (but in a smart way) often results in MP3 being preferred over the original.
This is a good example of how 'removal' of info can result in an improved SQ.
Also, like with filters as used in the upcoming amp, I can remove information by removing a peak and thereby increasing the SQ.
Of course in both cases while SQ increases by removal of info this is not what your typical amp can/will do.
I do 'understand' the thought from purists that state 'every added component in the chain removes something and thus less is more'.
Complete bogus of course but understandable thoughts... If only they realised how MANY components (and often even FAR from audiophile grade) the signal had passed before it was recorded is staggering.



Actually it CAN be a case of merely increasing the signal level, even an increase of 0.5dB will be perceived as better SQ yet not perceived as louder.
Also the point Ian touched about dynamic range (reaching clipping levels just slightly) can evoke this increase in SQ.



See possible reasons mentioned above.

The Karmeleon amp does the same as a C'Moy (in terms of providing a lighter load for the source but has more 'power' behind it and above all corrects what is missing or is 'too much' in the headphone. When used as a 'flat' amp the only thing it can do better than a C'Moy is having a MUCH higher output current capability.

That is a heavy subject.
Perhaps in the 'perception' area ?

Most people have a soundcard with audio in and output and operate on Windows they could install RMAA.
It would be interesting to see some spectrum and FR plots made with different tubes under the same circumstances (audio level).

This is pretty easy to set up and could clarify some found differences or debunk some.
I don't have that many different tubes myself.
I do know Jeremy did some measurements in the past that showed differences between tubes.

because the Ember is a non-feedback design the gain of the tube has a direct relation on the SPL.
This means that when a tube has a somewhat higher gain it may be preferred over another tube that has slightly less gain while having the same characteristics otherwise.
When the volume control hasn't been touched it is very easy to mistake a slightly louder tube as better sounding yet if you were to equalize the level before listening the differences may be non existent, providing FR is wide enough and the added harmonics are below the audible threshold (or masked by the music)

Jeremy has a lot of tubes but little time to play around.
He has tubes that showed roll-off in the lows and some with rolled off highs (this is potentially audible even in blind tests but of course already VERY measurable).
Noisier tubes can also affect perception.
I remember some tubes Jeremy tested having different harmonic contents.

In the tests I did I also found that some tubes had 1st harmonic at -40dB while a different tube had it around -60dB yet they had the same frequency response.
This was with the amp maxed out near clipping level.
Differences get smaller when smaller signals are present.
The -40dB harmonics would be audible.
Tubes do not have a linear increasing distortion which also adds to the discussion.

At lower (more realistic levels) distortion levels dropped down to -70dB and doubt this would be audible.

So yes there are differences between tubes in harmonic contents at considerable levels (when used as a tube buffer for instance or when driving headphones like HE-6) and also in FR between some tubes.
Most ECC88 derivatives though perform largely similar at normal listening levels.
t.b.h. I don't hear any differences between very similar tubes on normal listening levels myself (when tested blind and level matched) but can measure them.