Hi everyone, I know this post has been inactive for over a year, but since the OP's question hasn't been answered yet, and for all those like me who have come to this post with the same question, I decided to reply and share my experience.
At the end of 2024, a Washburn BD30B from the music school where I study got to my hands completely dead. I had heard it a few days before with a clipping sound, not from a saturated speaker but rather from an overloaded power supply, even with a passive bass and the volume set very low.
When I took it apart and checked the PCB, in addition to the transformer with a burnt primary winding, I found at first glance this R38 burned and almost illegible, except for the first green band.
After checking the transistors, I also found burnt Q34 (C1815, NPN) and Q33/Q35 (A1015, PNP), which I later replaced with BC337 and BC327 respectively, despite the pinout.
The power transistors were two 2DS2061 and were in good condition.
The transformer had lost its label, so judging by the power supply filtering capacitors (2200uF 36V), I replaced it with an 18V+18V 1A transformer.
I took the time to draw the schematic of the PCB power stage to simulate it and determine the appropriate value of R38 to achieve linear zone biasing of the Darlington pair.
At least in the simulation, with completely generic transistor models, I was able to achieve a current of 6 mA through R38 and 64 mA of bias current through the 2DS2061, with about 500 mV at the speaker output.
Based on trial and error in the simulation, it seemed that the value of R38 could vary within a range between 1 ohm and 240 ohms, maintaining a bias current of 60 ~ 70 mA. Values above 240 ohms increased the bias current to over 100 mA and more, that could lead to unnecessary dissipation with idle input.
Long story short, I replaced R38 with a 50 ohm (two 100 ohms in parallel) since I couldn't get a 56 ohm one, following the assumption of the first green band.
On the PCB, the measured current of R38 resulted in values very close to the simulation (approximately 6 mA), but the darlington bias current was much lower (about 10 mA), with an idle input speaker output voltage of about 110 mV (closer to 0 V than the simulation).
The amplifier sounded clean even at high volume settings, although it probably had imperceptible harmonics due to the lack of fine tuning of the bias current. It's back to music school to give (I hope) a few more hours of students practice.
I hope this analysis, the schematic, and even more so the simulation file (created with SimulIDE-1.1.0 under Ubuntu, .sim1 extension renamed to .sch) will be useful to others who stumble with a fault in one of these amplifiers.
Regards,
Pablo