I did some research on PCB necking , the idea being to make a controlled fusing track section,
and also for using traces as current sense resistors.
See page 6 of the ZXCT1010 (current sense amp) datasheet
http://www.diodes.com/_files/datasheets/ZXCT1010.pdfand
http://www.diodes.com/_files/datasheets/ZXCT1009.pdfYou can also save $$ by using BCV61 or BCV62 as the sense amplifiers.
The copper has a large tempco, so useless for accurate metering, but OK for overload protection where a higher current limit is acceptable at lower temperatures.
If you take a short section of (say 8mil) track and use it to join two large areas of copper together then it can carry significantly more current than a long section of trace.
From memory it was a rule of 3's , Say for example a trace is 10mil wide, then if you use 3 times this length (30mil) and join it to fills (pads) 3 times as wide e.g. 30mil , that were at least 3 x trace long (30mil) then you would double the trace current capacity.
So you could make a serpentine shunt, and put blobs at every turn.
I generally found any kind of PCB trace shunt impractical for generating 100mV type signals.
But 1000mils of 100mil (in 1oz) with a gain of 100 amplifier was pretty good for detecting current in the 5 to 10A range
These shunts are used for chopper MOSFET protection on actuator drivers , these might be 4A at 12v nominal load, but should be able to ride through 8A surges and go into protection mode above 10A or a short circuit.
I've actually gone away from the shunt approach now, and instead measure the VDS of the MOSFET, including a PTC thermally bonded to the drain pad, this provides overtemperature, underdrive, overcurrent protection , and the PTC provides failsafe protection; I'm measuring the VDS using a 2N7002 in semiparallel to the big MOSFET, surprisingly simple, yet accurate, the VDS measured is large enough to measure directly with a CPU, (4A = about 100mV = 50bits on ADC)