Alan8947
Full Member level 4
120W of heat!? Then you are making a class-A home heater, not a modern audio amplifier that has a small class-AB idle current and lots of negative feedback for extremely low distortion. The negative feedback and wide bandwidth does the error correction like most modern hifi amps.
Crossover distortion is a totally different animal. It is constant and is independent to the input signal. That is it affect the program the most when the signal is low. Another miss conception in THD spec that is tested in large signal. Crossover distortion usually contain higher harmonics, You cannot just use GNFB to lower crossover distortion as loopgain ( or they called feedback gain in audio world) is very low above 10KHz. For example, even if you have BW of 200KHz, at 20KHz, loopgain is only 20dB. You don't cancel much with GNFB.
I studied hundreds of schematics of all different kind of hifi amps, from cheamp Yamaha to Krell, Mark Levingson, Bryston etc. None has very high loop gain. They are cut corners in the designs like I said before. The more expensive ones just use more output transistor pairs to cover up the short comings ( and it really works, that's the key for lowering crossover distortion and lower output impedance). In fact, it is not a good practice at all to rely on GNFB to lower distortion. GNFB should be used as the final touch.
I use 1A of idle current to get a bigger class A region. Higher idle current shift the crossover region to higher signal level......that is signal to THD ratio is much bigger. for 1A, the amp remains in class A until the peak signal is 2A X 8ohm = 16V for 8ohm speaker or W= 1/2 X 16V X 2A = 16W. At this level, crossover distortion is drown out.
I don't know of any other way around this except designing a TRUE error correction local feedback around the OPS. That's my next design.
The only other design has some hope is class H that has a small class A amp as the main amp, then modulate the two rails according to the input signal to keep the voltage across the middle amp low to reduce heat. BUT this has a big short coming both in implementation and theory. In theory, modulation of the rail will introduce distortion to the main amp as the collector output impedance is NOT infinite. That's it has finite Early Voltage. Secondary and more problematic is with collectors of output transistors not bypassed solidly to ground, you invite parasitic oscillation.
So many amps have instability, I really don't think people appreciate the difficulty of stabilizing these audio amps. Most just use an inductor to separate the output that drive the speaker from the output of the power amp. This is really cheating and has big consequence in THD. Another miss conception in THD spec that using a linear resistor load. The inductor does NOT increase THD measure BUT become a problem when driving a real speaker................... I ranting on to much already.