High Current Issue with Complementary 300W Amplifier

manish_sharma_89

Sep 16, 2026
1
Joined
Sep 16, 2026
Messages
1
I am working on a DIY 300 W audio power amplifier (A300) based on a complementary VAS architecture with a differential input stage, complementary VAS, bias/driver stage, and multiple parallel output transistors (NJW0281G/NJW0302G).

The amplifier works correctly in LTspice simulation. With the intended ±75 V supply, the simulated quiescent current is only around 10–11 mA.

However, during initial real-world testing at reduced supply voltage, I am getting a very high DC current draw:

I have used a transformer of 66 0 66 then i have a power supply
  • 6 V → approximately ±5 V rails, but about 2.2 A DC
  • 12 V → approximately ±7 V rails, but about 5 A DC
  • Expected from simulation at these conditions: approximately ±8.5 V / 10 mA and ±17 V / 11 mA, respectively.
The output transistors and other devices heat up very quickly even at this low supply voltage. The power supply and transformer appear to be capable of supplying the current, but the amplifier itself is drawing far more current than expected.

I have checked the schematic against the simulation and am currently investigating the bias voltage, transistor pinouts/orientation, driver/output-stage wiring, emitter resistors, grounding, compensation, layout design rules.

One question I would particularly appreciate help with is: what are the most likely causes of several amps of quiescent current in a complementary-VAS amplifier when the simulation shows only ~10 mA? Also, what measurements would you recommend making first to isolate whether the problem is in the bias/driver stage or the output stage?

For more reference i can send the schematic as well which is working well in simulation in lt spice.
 

Martaine2005

May 12, 2015
5,314
Joined
May 12, 2015
Messages
5,314
Measure the voltage drop across the emitter resistors. Should be very low, for example 10mV.
 

AnalogKid

Jun 10, 2015
3,055
Joined
Jun 10, 2015
Messages
3,055
The sim results are based on perfect components that are perfectly matched. The real world has neither.

ak
 
Top