Styx said:
BJT can handle alot more current and have a lower on-state conduction losses.
Not at all. It's difficult to compare since you there are large and small BJTs and FETs, so it's not fair to pick two out of a catalog and compare them.
There are many MOSFETs capable of handling quite high currents.
On the whole, a BJT will consume more power in the on-state. It cannot switch with less than a 0.3v voltage drop, and it takes a lot of base current to do it. Now there certainly are MOSFETs with poor rDS-on that they exceed a 0.3v drop under load, but that's hardly a rule. I commonly use MOSFETs with an rDS-on on 0.02 ohms or less.
The bipolar's emitter current can be problematic. If you want an on-state voltage drop of 0.3v at 5 amps, you might need to give it 200mA of base current (gain is poor as you get into the overdrive region). Say it's a motor and you wanted to meet that spec so it won't overheat the transistor or stall the motor when heavily loaded. But most of the time your motors only draw 500mA. Well, unless you're doing something funky with the driver, you must end up putting in an emitter resistor that always provides 200mA to the base no matter what the load. You can't really adjust the current to only the magnitude needed to drive the load at the time.
If you have a 12V supply and a 5V reg, you're sucking 200mA off the 12V supply, or 2.4W, to drive a 500mA load. In practice, you'd choose a more practical base current and live with the added voltage drop & heat in the BJT. Or the far more practical solution is to use a FET. A FET would take no current on its gate under high or low load and there is no prob finding one with a voltage drop <0.05v under a 5 amp load. There are some awesome switching MOSFETs in SO-8 pkgs that handle high currents without generating significant heat due to such low rDS-on.
MOSFETs can easily be placed in parallel, bipolars cannot unless external emitter resistors are added. The external resistance generates additional efficiency and voltage drop losses.