Scott:
A buffer to drive the MOSFET is certainly most likely necessary. A standard 8051 type output pin will be unsatisfactory for trying to drive the gate of an N-Channel MOSFET that exhibits a full ON threshold voltage at or over the processor V_DD. The standard type 8051 type output will also be unsatisfactory for producing a gate drive to the MOSFET if the gate capacitance is larger than say about 100 picofarads and even at that the turn on will not be a sharp transition.

It can be very handy to use the type of driver chip that has the built-in charge pump to produce a higher than V_DD gate drive voltage, if your MOSFET requires this. This can save the need to generate an additional power supply voltage. The driver chips of this type and/or those like the Micrel parts you mentioned tend to be used in applications where there is a need to have very fast and firm ON and OFF switching of the MOSFET. This is important when there is a need to improve efficiency by eliminating heat loss in the MOSFET. And also this becomes much more important in applications where the switching activity has a high frequency.

It sounds like your application is a very low duty cycle one where switching speed and/or the loss of a little power in the MOSFET would be of little concern. I suspect that you could realize a very effective driver for an N-channel MOSFET by using a PNP transistor to pull the gate to +12 volts for the ON condition and then bias the part OFF with a suitable resistor to ground from the gate (Noting of course that while the PNP transistor ON it will be also sourcing current into this gate bias resistor). The PNP transistor can then be driven by another small N channel logic FET, NPN transistor, or opto coupler output.

The need for opto isolation will depend entirely on how the load of the solenoid, the grounds and the corresponding 50 volt power source relate back to your microcontroller setup. For example, if the 50 volt supply is subregulated to the V_DD for your processor and the DC grounds are common then the use of the opto couplers will be of some questionable value because you would still need to very carefully consider your grounding system anyway. On the other hand if the 50 volt supply is a completely different supply from that used to power the microcontroller then it can be highly advantageous to separate the grounds and use opto isolation. I can say that I have seen a lot of design recommendation noise over the years wherein optoisolation is promoted and yet its utility is not even worth the cost or the effort. Everything depends upon your system requirements.

Michael Karas