amplifier - Problems in push-pull biasing - Electrical Engineering Stack Exchange

i'm trying make push-pull audio amplifier pair of power darlingtons (tip142 , tip147), i'm having problems biasing.

here's diagram of 1 of configurations:

here i'm using 3 diodes, used different things in place try correct biasing voltage.

with biasing voltage lower 2.2 v (between bases of transistors) has crossover distortion, because there'll small parts of signal none conducting.

with 2.2 v or more, both conducting , there's no more distortion, there's problem: current flowing between them escalates heat until take on fire (being starting current isn't great , they're on large heatsink, heatsink doesn't have time heat, fast happens). they're stable biasing voltages lower 2.1 v, there's distortion.

the components heating here output transistors, i've cheked , none of resistors or diodes heating up.

i've checked if negative feedback help, didn't, because voltage in output doesn't change current flowing through them escalates (both conducting more , more equally, if increasing biasing voltage).

also, if note on base of transistor driving tip's, you'll notice 47k potentiometer use make sure both conducting equally when there's no signal (so output rest @ 0 v).

but somewhy, point there's more distortion. if make 1 of them conducting little bit more other (in way output not rest @ 0v in 1.2 v) there'll no or less distortion.

in simulations in multisim none of problems occur, shows no distortion when it's biased, in reality have either distortion or great instability.

i know put 2 resistors between emmiters limit current, load (my speaker) 4 ohms, 1 ohm resistors create great loss.

so, can me?

thanks!

emitter resistors on darlingtons "soften" biasing problem. 1 ohm have found - consume of available power.

biasing darlingtons class ab tricky. if find 2.8 volt battery replace series diode string, darlingtons @ risk of thermal runaway.
q3 (below) along d1, d2 , variable resistor r3 form almost-constant voltage source of 2.8v. can vary voltage variable resistor.
counteract thermal runaway problem, q3 can thermally tied heat sink q1, q2 sit.

^{simulate circuit – schematic created using circuitlab}