The circuit I sent you is not finished yet. Yesterday, I connected it
to the output stage to test how it behaves. At first, it oscillated at
a frequency of 660kHz. I realised this because the heatsink started
getting steadily hotter. To decouple high ultrasonic frequencies I
used a decoupling capacitor betweent the base and collector of the
transistor taking the output from the differential pair. After
energizing again the circuit, the high frequency was gone, but I am
still getting the first and second harmonics of the power supply's
ripple. This means, the power supply's ripple is being injected into
the amplifier's signal. The most obvious cause for this to happen, is
a bias voltage that is slightly dependent on the instantaneous supply
voltage. I think, I will soon have that also resolved. I will connect
a resistance and an electrolytic capacitor across the current source's
bias diodes. Since current sources have extremely high impedances,
while allowing appreciable current to pass, a slight variation in
biasing them, can result in getting harmonics from the supply's
ripple.