Rod Elliott has some very interesting thoughts about intermodulation distortion. Basically it requires asymmetry, either in the nonlinearity or the signal. He then kind of dismisses his own idea with the quip that music is asymmetrical anyway.
But as I was just thinking, the asymmetry is only partial. There is also partial symmetry, so in fact his observation does relate to music reproduction...in a major part.
The other strange thing is that supposedly good sounding even harmonics are precisely the asymmetrical ones. Or more precisely, asymetrical non-linearity is what gives rise to even harmonics, and symmetrical non-linearities give rise to odd ones. So the supposedly good sounding even harmonics are the ones in greater part contributing to the overall IM distortion, in music.
So basically this comes down to, you can't hand wave or excuse distortion away, as a wide swath of serious audiophiles do, with SET, NOS, zero feedback, whatever the rationale, if it leads to significant distortion, that is not good. (This is not to say, that you might always claim some benefit of greater need, but it has to overcome a significant loss...not a nothing...if harmonic distortion AND IM are added.)
And my own personal threshold of good performance I've noted as 0.01%, however, that should also be weighted by the audibility of harmonics, irritability of the harmonics, and then also rated for contribution to IM.
I was going to say that the requirement for a even-harmonic cancelling balanced device like the Master 7 Dac should be made more stringent because of relative audibility and irritation of odd harmonics left over after the cancellation. However, when IM is considered...and pure balanced operation cancels asymmetrical linearity generally, and hence the IM associated with it as well, perhaps one should not be so critical as an odd-mostly distortion spectrum such as the one so produced, as long as it is low enough. And the same applies to all true balanced operating devices and perhaps nearly as much for quasi balanced arrangements like push pull. These arrangements tend to be beneficial, despite the odd mostly distortion spectrum ultimately produced at low levels. Of course the best thing is...linearity all the way.
Not that I'm claiming odds are good per se, but I did see one poster claim that the 5th harmonic adds "sweetness" like musical fifths.
I seem to recall some higher ones, and especially combinations of ones, can be quite nasty.
But there's nothing good about asymmetrical nonlinearity either.
Elliott points out that traditional IM measures barely capture the possibilities of IM.
But as I was just thinking, the asymmetry is only partial. There is also partial symmetry, so in fact his observation does relate to music reproduction...in a major part.
The other strange thing is that supposedly good sounding even harmonics are precisely the asymmetrical ones. Or more precisely, asymetrical non-linearity is what gives rise to even harmonics, and symmetrical non-linearities give rise to odd ones. So the supposedly good sounding even harmonics are the ones in greater part contributing to the overall IM distortion, in music.
So basically this comes down to, you can't hand wave or excuse distortion away, as a wide swath of serious audiophiles do, with SET, NOS, zero feedback, whatever the rationale, if it leads to significant distortion, that is not good. (This is not to say, that you might always claim some benefit of greater need, but it has to overcome a significant loss...not a nothing...if harmonic distortion AND IM are added.)
And my own personal threshold of good performance I've noted as 0.01%, however, that should also be weighted by the audibility of harmonics, irritability of the harmonics, and then also rated for contribution to IM.
I was going to say that the requirement for a even-harmonic cancelling balanced device like the Master 7 Dac should be made more stringent because of relative audibility and irritation of odd harmonics left over after the cancellation. However, when IM is considered...and pure balanced operation cancels asymmetrical linearity generally, and hence the IM associated with it as well, perhaps one should not be so critical as an odd-mostly distortion spectrum such as the one so produced, as long as it is low enough. And the same applies to all true balanced operating devices and perhaps nearly as much for quasi balanced arrangements like push pull. These arrangements tend to be beneficial, despite the odd mostly distortion spectrum ultimately produced at low levels. Of course the best thing is...linearity all the way.
Not that I'm claiming odds are good per se, but I did see one poster claim that the 5th harmonic adds "sweetness" like musical fifths.
I seem to recall some higher ones, and especially combinations of ones, can be quite nasty.
But there's nothing good about asymmetrical nonlinearity either.
Elliott points out that traditional IM measures barely capture the possibilities of IM.
No comments:
Post a Comment