I find some precision SMD resistors are 4-terminal, like vishay's CSM3637P. It's constructed as below:
It seems it only make a slot on each pad, then get two extra pad to do the voltage measurement.
If using a normal two terminals chip resistor, I can route it as below, then what's the benefit using the 4-terminal one?
In your board layout, the voltage terminals are sensing the resistance made up of the resistor component, the solder down to the board, and an ill-defined short length of copper of odd geometry.
If the resistor was a few ohms, you could probably neglect the solder and the copper. With a few milli-ohms, it is almost certainly significant.
The solder joint may vary from run to run, and will vary if you change the pad size. Any time you change the pad size or geometry, the copper factor may change slightly. You would have to characterise these factors yourself, and repeat that any time you changed something.
However, if you use a 4 terminal resistor, then the resistor manufacturer has measured his resistor in the way you are going to connect it, and is guarranteeing the resistance in that configuration. You can change pad size, geometry, solder conditions, and your measured 4 terminal resistance will stay the same.
That is the whole point of the 4 terminal connection, repeatability in the face of variable connections to the device itself.
These resistors are normally used for measuring quite high current. In your 2-pad layout all of that current has to flow through the 2 solder joints, which will drop voltage and therefore contribute an error to your measurement. By having 4 pads the measurement pads do not pass high current, do not drop voltage, and so do not introduce and error in the measurement.
