Since these resistors are very low-valued, the connecting wire’s resistance will be a significant percentage of the resistor value. In the schematic, these are A and D. These are called the “force” wires and are the pair of large pads in your photo.
To prevent reading the spurious resistance, a pair of “sense” wires are connected immediately at the resistor’s resistive element. These are labeled B and C and are the pair of small pads. As also stated in the schematic, the meter connected to these pads will have an extremely large impedance, and therefore will not impose a loading on B and C.

 

shunt is a shunt... just a resistor. this particular version has 4 terminals as shown in the image. A,D is used for high current. B,C is used for measuring voltage drop.



One could also use regular 2-terminal shunt by making own Kelvin connections. This is one of several high current channels I just designed yesterday:

the top and bottom red areas are copper pour (zone) for high current. thin tracks in the middle are for sensing.

 

OK, yeah, I get it now. Thanks.

 

Excellent thread! The sense pads were indeed shorted to the force pads. This is why the shunt wasn't delivering it's rated specification. This condition will be rectified shortly (pun unintended). The current source is "hard". The feedback resistor for this current source is not in the path. This kelvin shunt is only used to verify/confirm/calibrate/ensure that current source is set perfectly (100mV for 1 amp, 300mV for three, or whatever). The current source is stuffing current across a thin film to measure the voltage drop , which is plucked-off with pogo pins over a set distance. Having confidence the current is set to a fixed value every morning assures that the film readings are relatable to a known quantity. Using a new Fluke 87v to double check was my way of reassuring myself, and then, un-assuring myself
I also wanted to increase current to 9A, to created greater signal in the mV range, and was unable to use this calibration resister, because it is only rated for 3W. Thus, I began test-driving my new Fluke around the circuit, and that's when I found the discrepancy.

 

K meter probes