> how transistors Q3, Q7 and Q10 work...
They are not transistors. They are two-cent low-spec Zeners.
Look at, fer example, OA2 Q3 Q5. OA2 is powered from +24V and 0V, its output can't be trusted to swing any more than +2V to +22V. The base of Q5 sits 0.6V to 1.2V below the +24V rail, or at +22.8V to +23.4V. The opamp output can't swing high enough to run Q5 at any less than WIDE OPEN.
Q3 looks like a transistor, but only two pins are used. You can treat it as two diodes in series. Both diode arrows point to the Base pin. The C-B diode is forward biased, drops 0.6V. The B-E diode is reverse biased so at first it does not conduct. But feedback drives OA2 output negative trying to force Q5 to the On condition (otherwise R12 holds Q5 off). And the reverse breakdown voltage of the Emitter junction of most transistors is 7V (as opposed to the Collector junction that can stand 40V or more). So OA2 output pulls down 0.6V+0.6V+7V= 8.2V down from the +24V rail, the Zener/transistor breaks down and conducts good, flows current to R12 and Q5 Base, turns Q5 On. OA2 output is at ~+16V, or 8V below its positive supply pin, well within its ability.
They could have used opposite polarity on Q5 and avoided the Q3 Zener hack (and overcompensating that LM301). But the dropout voltage would be lower. Using a NPN for Q5 as an emitter follower, and assuming 2V drop in OA2 and ~1V drop B-E in Q5, the dropout voltage is at least 3V. With the PNP and the Zener-hack, it may be less than 1V.
They could have actually bought Zeners. But with Zeners you pay for Known Voltage and some other specs. Here, we don't really care what the voltage is: could be 4V or 10V, the opamp would still find a balance point. We don't care if the breakdown knee is sharp or soft: OA2 and Q5 have plenty of gain to spare. And DBX was probably buying crates of transistors and Selecting-out precision-match pairs for the VCAs (at least in my box, the VCA is discrete) and various utility functions.... the leftovers with odd Vbe and low Hfe can still be fine Zeners, and no extra part-number to order and keep in inventory.
BTW, "OA1" looooks like LM723.
They are not transistors. They are two-cent low-spec Zeners.
Look at, fer example, OA2 Q3 Q5. OA2 is powered from +24V and 0V, its output can't be trusted to swing any more than +2V to +22V. The base of Q5 sits 0.6V to 1.2V below the +24V rail, or at +22.8V to +23.4V. The opamp output can't swing high enough to run Q5 at any less than WIDE OPEN.
Q3 looks like a transistor, but only two pins are used. You can treat it as two diodes in series. Both diode arrows point to the Base pin. The C-B diode is forward biased, drops 0.6V. The B-E diode is reverse biased so at first it does not conduct. But feedback drives OA2 output negative trying to force Q5 to the On condition (otherwise R12 holds Q5 off). And the reverse breakdown voltage of the Emitter junction of most transistors is 7V (as opposed to the Collector junction that can stand 40V or more). So OA2 output pulls down 0.6V+0.6V+7V= 8.2V down from the +24V rail, the Zener/transistor breaks down and conducts good, flows current to R12 and Q5 Base, turns Q5 On. OA2 output is at ~+16V, or 8V below its positive supply pin, well within its ability.
They could have used opposite polarity on Q5 and avoided the Q3 Zener hack (and overcompensating that LM301). But the dropout voltage would be lower. Using a NPN for Q5 as an emitter follower, and assuming 2V drop in OA2 and ~1V drop B-E in Q5, the dropout voltage is at least 3V. With the PNP and the Zener-hack, it may be less than 1V.
They could have actually bought Zeners. But with Zeners you pay for Known Voltage and some other specs. Here, we don't really care what the voltage is: could be 4V or 10V, the opamp would still find a balance point. We don't care if the breakdown knee is sharp or soft: OA2 and Q5 have plenty of gain to spare. And DBX was probably buying crates of transistors and Selecting-out precision-match pairs for the VCAs (at least in my box, the VCA is discrete) and various utility functions.... the leftovers with odd Vbe and low Hfe can still be fine Zeners, and no extra part-number to order and keep in inventory.
BTW, "OA1" looooks like LM723.