Originally Posted by Matt Syson
On the basis that power supplies have a finite value, everything will clip at some point unless some form of 'clamping' is installed.
Back to back zener diodes could be a cheap and dirty method, or various other permutations. Not saying it would sound better than fast recovery from occasional peaks though.
The SSM2015 as a mic amp would burst into a short but constant period of VHF oscillation if 'triggered' by a peak. It didn't seem to matter how long the 'trigger' was.
Clamping is pretty much controlled clipping.
it is a generally unavoidable fact that when an output stage saturates any overall negative feedback is no longer closing the loop, so internal circuitry will keep trying to drive the output further. Multiple internal clamps are routinely used to prevent these internal nodes from getting too far out of whack and requiring a bunch of return to return to stable operation after the overload event is removed.
Clamps in the NF to prevent saturating the device may be useful, if the IC design is not already managing it's recovery time. Generally easier to do complex anti-sat circuits inside an IC than externally with discrete design.
Zeners in the feedback have their own issues, and I suspect a very high speed part, like JW seems to favor, may recover faster than a slow zener.
Note: Instability around clipping could be evidence of inadequate power supply decoupling, or not...ymmv