Thanks, the credit goes to NP F5. Still it is also very close to basic SSA, somehow logical step to a really low parts count, still maintaining input BJT-s and to have bias/offset setting. I couldn't resist to draw it ...
1K trimmer is replaced by a 2k2 fixed and in the negative part 1k + 2k trimmer.
RC top and bottom is set to 2.2k=> more OLG.
BC550 and 560 is replaced by KSC3503 and KSA1381.
at 24V it will settle to about 1.2 - 1.4 amp.
for an equal current in the mosfet an vgs difference on the 10n20 and 10p20 from exicon of about ~1V. This will lead to and offset. Maybe a diode in series with one of the collector resistors will help a lot!?.