< Rupert Neve used to have a simple, though not scientifically controlled,
demonstration, which he used as his justification for extended bandwidth needs.
He would play a sine wave at the upper end of a person's perception, 15kHz or
maybe higher. The respondent would agree that it was audible. Then Neve played
a square wave at that same frequency. The harmonics in the square wave would
obviously all be entirely beyond ones supposed 20kHz limit to perceive. Yet
invariably the respondents would be able to identify the square wave as audibly
distinct from the sine wave, even though theory would tell us that only the
fundamental (a sine wave) of the square wave should be audible. Neve arrived at
this test after an engineer told him that one module on a Neve desk was
malfunctioning compared to the others. The engineer could hear a difference in
that channel, & nobody else could. Tests revealed that that channel was
oscillating at something like 60kHz. Perhaps it's audio mythology, but having
come from the mouth of Rupert Neve I'm inclined to give it credence.>
My explanation of these effects would be that the equipment used for these
tests (most likely the power amplifier or speaker) introduced intermodulation
distortion due to the unavoidable nonlinearities. The steep slopes of a 15 kHz
square wave have a very broad ultrasonic frequency spectrum. These different
ultrasonic components can then modulate each other and the difference between
at least two of those components would be audible. These effects can simply be
prevented by limiting the bandwidth of the power amplifier to 20 kHz. If the
power amplifier and speaker were perfectly linear, I guess there would still be
no audible difference.
Regards,
Raimund
|