MOS transistors have three terminals (G, D, S). Two of them should be used as input and output ports, and the remaining one terminal is grounded. Depending on which terminal is grounded, three amplifier configurations can be derived.
Common source:
+ input: G
+ output: D
+ ground: S
Common gate:
+ input: S
+ output: D
+ ground: G
Common drain:
+ input: G
+ output: S
+ ground: D

As you can see the input is always G or S never D. Is there a reason for this?
Why the configuration with input at D never exists?

 

Its for impedance matching and voltage gain, in common base or gate, the input is the emitter or source, used a lot in rf front end amplifiers for high input impedance and the gain is usually 1.

 

As you can see the input is always G or S never D. Is there a reason for this?
Why the configuration with input at D never exists?

At a simple, practical level, the reason is that such a circuit simply doesn't do anything useful-- there is no voltage gain, no current gain, no power gain. And in a logic circuit context, no usable switching action occurs. The same applies to any circuit configuration which attempts to use the gate as an output; it just doesn't work. BJTs are no different; we don't use the collector as an input, nor the base as an output.

You could arrive at the same conclusion analytically by constructing such a circuit (on paper, of course) and deriving its transfer function from the device characteristics along with the component values in the circuit: it just doesn't work.

 

If an N-type MOSFET is used as a high-side switch for a load, the +V supply would be input at the drain.

 

In a FET the drain and source can be generally interchanged (ignoring any substrate diodes).
In that case the terminals that become the source and drain are determined by the relative bias voltages on the three terminals.
For example in CMOS IC circuits there is no physical difference between the source and drain connections as compared to the gate channel.

This characteristic is seen in power MOSFETs when used in bridge rectifier circuits, as synchronous flyback diodes in switching power supplies, or when used as a reverse-power block (diode substitute).
In each of those applications the normal drain-source current flow is actually the opposite of what's considered normal (i.e. the current flows in the forward direction of the substrate diode).

 

I would consider the 'input' of a MOSFET current source to be the drain.

 

I would consider the 'input' of a MOSFET current source to be the drain.

Don't understand that.
I consider an "input" something that controls the device and the drain does not do that.
The drain carries the constant current that's being controlled, independent of the drain voltage.