daArry,
Some more small error,
The R10 shall be 47 Kohm not 47 ohm, and as the guy`s already have say, the +in transistor collector shall go direct to the T1 base, not via C3.
You can also take away R9, R11 and one of the two diodes pair, so all three constant current source/CCS transistors base share R10 and two diodes in series, so you get a more simple test circuit to start with.
If you will test with all this CCS for the input pair, common emitter and emitter follower transistors, it is very important that you can measure the current this CCS shall draw through the transistors, so you know what you have in each transistor section.
The current mirror and cascade on the input pair is very good for many parameters, but it can be more easy to get your first discrete op-amp stable undertime you experiment with different current and gain in the transistors, use also a 470 ohm to 1 kohm resistor in series with your oscilloscope before you connect it to the output, because a scope cable can get your amp very instable. (if you not have a X1/X10 attenuation probe)
Tamas,
you are absolute right, it is a complex interplay between all this small compensations capasitors, because it depend on the pc-layout, transistor parameters and many other things, but what I mean, it can be more easy to start with out this capacitors and see what is happen from the beginning and then do a cut and try and see which capasitor or capasitors in different value range that will do the best job for the specific instable problem you have.
I like this discussion,
more amp design talk please.
It is fun to share experience,
but we need more PRR fact, he can his homework about the solid state basic rules.
--Bo