The emitter resistors provide local negative feedback to force sharing amongst the 5 output devices that are paralleled. Bipolar transistors exhibit a negative temperature coefficient on the base-emitter forward junction voltage. Thus if an identical voltage is applied to all 5 base connections (as exists in this design) the transistor with the lowest Vbe will turn on before the other 4 transistors and hog the base and collector current. If this is allowed to happen, this device will heat up more than the remaining 4 devices. Due to the negative tempco of the base-emitter junction, the Vbe will reduce even further, leading to yet more hogging. This description is the depiction of thermal runaway in bipolar transistor devices that are paralleled without emitter resistors.
If the emitter resistors are added to each of the paralleled transistors, this same transistor that was hogging the current before receives local negative feedback from its emitter resistor. The more current that flows in the base and collector of the hogging device, the larger the voltage that is built up in its emitter resistor, effectively reducing the voltage across its base-emitter junction and in turn reducing this transistors collector current. This example is what goes on in each of the 5 paralleled transistor devices.
The emitter resistors serve to keep thermal runaway of any one device in check and to force relatively equal sharing amongst the 5 devices. Sharing accuracy, within reason, is not a critical parameter unless you intend to eke out the last bit of performance and safe operating area in each of the 5 shared transistors. If you intend to push it to the limit, it is best to bench grade each transistor and select devices that are very closely matched for Vbe at a selected base current. This is a tedious process that requires good instrumentation and time (to temperature stabilize each device under test). Once you go to this trouble, you have to put precision resistors in the emitter locations (and possibly reduce their value as well) to gain the benefit of your matching work.
You will not hear the difference resulting from all this work, but you will be able to safely eke more power out of the same paralleled devices. That old adage, a chain is only as strong as its weakest link.
Hope this help the understanding of output stage balancing and what those resistors are for.
If the emitter resistors are added to each of the paralleled transistors, this same transistor that was hogging the current before receives local negative feedback from its emitter resistor. The more current that flows in the base and collector of the hogging device, the larger the voltage that is built up in its emitter resistor, effectively reducing the voltage across its base-emitter junction and in turn reducing this transistors collector current. This example is what goes on in each of the 5 paralleled transistor devices.
The emitter resistors serve to keep thermal runaway of any one device in check and to force relatively equal sharing amongst the 5 devices. Sharing accuracy, within reason, is not a critical parameter unless you intend to eke out the last bit of performance and safe operating area in each of the 5 shared transistors. If you intend to push it to the limit, it is best to bench grade each transistor and select devices that are very closely matched for Vbe at a selected base current. This is a tedious process that requires good instrumentation and time (to temperature stabilize each device under test). Once you go to this trouble, you have to put precision resistors in the emitter locations (and possibly reduce their value as well) to gain the benefit of your matching work.
You will not hear the difference resulting from all this work, but you will be able to safely eke more power out of the same paralleled devices. That old adage, a chain is only as strong as its weakest link.
Hope this help the understanding of output stage balancing and what those resistors are for.