Many amplifiers have several different stages where the sum of the voltage of the stages is between the supply rails of the amplifier. For example, for an input stage, there may be a constant current stage, a long tailed pair, a cascode stage, and a current mirror load. The VAS stage might have a combination of common-emitter, bias servo, and constant current stages. See the attached schematic PDF for an example of this. Since the voltage drop is shared among the transistors, does the VCE voltage rating of the transistors individually all need to handle the rail-to-rail voltage? I am concerned that during power-up or power-down or while establishing steady state biases that one of the transistors might enter CE breakdown temporarily and be damaged. There is the possibility of current inrush also which might cause temporary bouncing of the supply rails as well. What kind of safety margin should be used for VCE when the sum of the collector-emitter voltages add up to the rail-to-rail voltage drop?
Common practice is to choose transistors whose VCEmax on the datasheet, exceeds (|Vposrail| + |Vnegrail|). It's possible, it's not all that difficult to find them, and it's very safe.
The only exception in practice, is devices "under a cascode", i.e. whose VC or whose VCE is actively limited by a cascoding device which sets the collector voltage. Typically that's either a low-VDSmax JFET, or a wickedly superbeta BJT like ZTX618 or 2N5089 which also has a low VCEmax.
But today we have the MJL3281A with a 260V VCEmax and 15 amp ICEmax and a wide safe operating area. If you're interested in recreating the bigboy amps of yesteryear, like James Bongiorno's "Ampzilla", they you'll be stacking old power transistors in series to stay within VCEmax and safe operating area limits.
The only exception in practice, is devices "under a cascode", i.e. whose VC or whose VCE is actively limited by a cascoding device which sets the collector voltage. Typically that's either a low-VDSmax JFET, or a wickedly superbeta BJT like ZTX618 or 2N5089 which also has a low VCEmax.
But today we have the MJL3281A with a 260V VCEmax and 15 amp ICEmax and a wide safe operating area. If you're interested in recreating the bigboy amps of yesteryear, like James Bongiorno's "Ampzilla", they you'll be stacking old power transistors in series to stay within VCEmax and safe operating area limits.
An input pair is held close to ground by the input network and the feedback network so
can be rated for a single rail voltage plus the maximum a preamp might chuck at it, so I'd
say a 80--100V transistor would be fine for 70 ot 80V rails in that situation. A VAS or its
CC load would need to be 150 to 200V for the same rail voltages.
Agree about cascode - one of the advantages of a cascode is the first device can be selected for high gain and low noise, without having to worry about voltage too.
can be rated for a single rail voltage plus the maximum a preamp might chuck at it, so I'd
say a 80--100V transistor would be fine for 70 ot 80V rails in that situation. A VAS or its
CC load would need to be 150 to 200V for the same rail voltages.
Agree about cascode - one of the advantages of a cascode is the first device can be selected for high gain and low noise, without having to worry about voltage too.