Yes, English is not my native, as you see, so i have some problems to describe my question.
Imagine the situation: We have a musical CD(1411kbps, pcm). We encoded it by mp3 codec (128kbps). That was the first compression. The mp3 codec has cut some frequencies, cut some fragments, which are inaudible for human's ears (anyhow codec "considers" so). Then we encoded this mp3 file by ATRAC. So it is the second compression. As ATRAC is not the same as mp3 codec, it has another principle of compression. So ATRAC may cut one more "inaudible" fragments. For example, mp3 codec cut 10 fragments + ATRAC cut 5 fragments more because of other principle. So after second conpression we have 15 cutout fragments versus 10 in case of one compression. I think, i described the effect of double compression visually enough.
But if we use high quality atrac codec (256 for example), the problem of double compression becomes not so sharp (critical). With a high probability mp3 codec (128 or 192) has cut all the fragments which ATRAC (256) was only going to cut, so it wouldn't cut anything (or cut a little) and we would get a sound quality equal to mp3 file. So the double compression effect don't take place.
And when I talked about "direct mp3 downloading on MD unit", i ment there is no double compression and the sound quality is equal to mp3 file, which was downloaded. That's why I said, that "in some degree it solves the problem of direct mp3 playing".
So, the question stays the same: am I wrong?