L20 AMP。use only two NJW0302G

The reason for the occurrence of crossover distortion is the different slope of the transmission characteristics of output complementary transistors. They can be reduced by current control of the output stage, but Miller's correction imposes restrictions on such a solution. If you want to get rid of crossover distortion, you will have to get rid of Miller correction.

the second problem is the thermal distortion of the output transistors, which leads to the formation of parasitic frequency modulation. the speed qualities of the same transistor with different crystal temperatures are not the same, and this is also very noticeable in the output transistors.

the solution here is to use the Lokanti topology, because the bulk of the signal amplification work moves from the output stages to the driver stages, which operate at lower currents and lower temperatures and are more symmetrical in characteristics with complementary performance than high-power output transistors ...
Thank you for your reply.
Although my views are different. But you can refer to it.
 
LJM L20SE Design Notes

LJM L20SE Design Notes (*)

1. The amplifier is built on a green PCB with approximately 110 x 50 X 40 mm (LxWxH).

2. The audio signal first passes R1/C1 which form a first-order input low-pass filter that keeps out unwanted radio frequencies. R2 provides a return path for the input bias current of the amplifier’s input stage (note that the value of R2 is set to equal that of R14 so that voltage drops across these two resistors balance out any DC offset). The signal then reaches transistor Q2, which forms a differential amplifier input stage with Q3. It is fed a constant current from the constant current source (CCS) Q8/Q1/R3. Q4 and Q5 cascode the differential stage. They keep the heat power loss low and provide for low input capacitance and high bandwidth. Towards the negative supply line follows a current mirror comprised of Q6/R8/Q7/R9. It guarantees for good symmetry, high openloop gain and low sensitivity against disturbances on the supply rail (PSRR).

3. Q10 forms the voltage amplifying stage (VAS) that works on the CCS-connected Q9 as load. Between Q10 and Q9 we find the biasing network Q11/R17/R18/C8 that generates and regulates the bias voltage for the output stage.

4. R14 and R15/C3 define the amplification factor. The large capacitor C3 improves offset accuracy by reducing the closed-loop gain to unity at DC. That spares the use of an active DC-servo stage and functions rather well under most circumstances.

5. C5 is connected across Q10’s collector and base as a compensation cap to limit Q10 gain at high frequencies (Miller Effect).

6. The driver transistors Q12/Q14 and Q13/15 Darlington drive the power transistors Q16/Q18 and Q17/Q19. The bases of Q14 and Q15 are connected together by R23.

7. The power transistors are connected to the speakers by means of the R24/R26 and R25/R27 damper-resistors.

8. At the output we find a R28/C11 Zobel network (the R28 10 Ohm approximates to the expected loudspeaker load impedance and the C11 capacitor is invariably 100 nF) to mitigate amplifier high frequency instability due to loudspeaker voice-coil inductive reactance.

9. The diodes D3 and D4 are “catcher-diodes”, i.e., they have a dual purpose: they protect from accidental power supply reverse polarity and they also protect the amplifier from loudspeaker inductive loads that can push energy back into the amplifier output, resulting in a possible rise of the output voltage above the supply rails and possible destruction of the output transistors due to polarity reversal. These diodes catch away the overvoltage to their maximum forward voltage (hence the term catcher-diode).


(*) Based on Calvin’s work: L12-2, powerful, good, low-cost - calvins-audio-pages


J.A.
I do not see any BE-discharge resistor at the output transistors to prevent cross-conduction. Very risky!
 
  • Like
Reactions: 1 user