I found Krell's patent on what is being discussed. It appears that rather than adjusting the bias among the different possible levels based on the input signal, the adjustment is performed based on sensing of the load current. Which explains the reference by Martin Colloms to "high-speed current-feedback circuitry." The patent goes on to indicate that bias increases necessitated by abrupt transients will be "jumped to," but subsequent decreases, if called for, will occur one level at a time, and slowly.
So George is correct that in the event of a large and abrupt transient there will be some very brief instant of time during which the amp will not be operating in class A. I have no opinion regarding the degree to which that may be audibly significant.
Regards,
-- Al
So George is correct that in the event of a large and abrupt transient there will be some very brief instant of time during which the amp will not be operating in class A. I have no opinion regarding the degree to which that may be audibly significant.
Regards,
-- Al