The goal is to find out how large the ballast resistors must be to prevent current runaway and to control "current hogging" to a given limit. Current hogging is a phenomenon with bipolar transistors when operated in parallel in which the hotter of the set draws more current than the cooler one(s) and so heats up even further. The fix is to put series resistors ("ballasts") in the emitter circuits (see the drawing), but I've not been able to find anything published that says how to calculate the value of the resistors.
The model is for two BJT transistors, with their collectors and bases commoned and their emitters summed through individual ballast resistors. You need to know what the process-variation-caused...
In a search to replace a modified SMSL headphone amp that I use at work, I decided to build a DOA SeNNator, a variant of the filtering headphone amp for the Sennheiser HD 650, published by Solderdude over at - diyaudioheaven along with measurements of the HD650's responses.
It's based on the "HD 650 SeNNator" and "SeNNator build tips" pdfs. In adapting the design, DOAs were used. An upgrade to the filtering caps from polyster to polypropylene. To the original circuit, circuits and design ideas from:
Sam Groner - +/-30mv DC offset adjust and DC bypass cap.
Layout idea lifted from Jeff Steiger's CAPI-2 Line Stage Amp, from Capi-gear.
DC filter from Collective Cases, reducing the numerous different electrolytic caps and upgrading the linear regs. from LM217/337 to LT1963/3015.
Minimized mechanical assembly, using the default PWBs available in Kicad, each channel and the DC filter/regulator boards are in a Euro card form factor...