The idea is to protect the transistors from destructive inductive pulses from the speakers. Tronhead speaks of connecting them from emitter to collector, which is probably easiest to add to an amp which does not already have this protection, but the most common way is from the output line to the collectors (or the voltage source, which amounts to the same thing on most amps).
Some Marantz amps use a 'common emitter' output arrangement, in which the collectors are connected to the output and the emitters to the voltage source, which is why specific directions like 'connect the anode of the diode to the emitter and the cathode to the collector' is misleading and will result in blown fuses and smoked diodes (at the least). The idea is captured here, with the diodes reverse-biased and connected to the output line, and is independent of the output stage topology.
Some Marantz amps use a 'common emitter' output arrangement, in which the collectors are connected to the output and the emitters to the voltage source, which is why specific directions like 'connect the anode of the diode to the emitter and the cathode to the collector' is misleading and will result in blown fuses and smoked diodes (at the least). The idea is captured here, with the diodes reverse-biased and connected to the output line, and is independent of the output stage topology.
