140416-0800 EDT
ptonsparky:
Since your goal is to sense an overload there are several possibilities, and you don't need any high accuracy.
Ideally you would measure power and not current. If you were not concerned with detection of phase loss on a three phase system, then a single phase sensor could be used.
If you use a current sensor (shunt, current transformer, or magnetic field sensor), then this sensor can have a full scale rating equal to something somewhat above your normal full load need so long as any overload conditions don't exceed the sensor's power dissipation capability. Suppose your normal full load was 200 A, then a 400 to 500 A transducer probably would be a good choice.
When current transformers are overloaded, then they simply start into magnetic saturation of the core and accuracy gradually decreases as saturation increases. A Hall type sensor has a much more square saturation characteristic. A simple resistor shunt is only a power dissipation problem.
There are some Hall current sensors with high accuracy that use magnetic feedback that have an inherent design defect that creates a problem when saturated, but this is not because of the basic Hall component.
Some websites:
http://www.wago.com/infomaterial/pdf/51285476.pdf very confusing.
http://fwbell.com/downloads/files/Methods_Current_Measurement.pdf I skimmed, well written.
http://fwbell.com/products/current-sensors-open-loop.aspx F.W.Bell now part of a British company. This web page is better than some of their others.
http://www.loadcontrols.com/products/products.html I have used their product many years ago and it worked very well. Still have the unit and use it once in a while. I don't like the website. It is not clear and adequately informative. I don't think that many websites are created by people that actual put themselves in the position of the user. For example I can converge on what I want with a paper copy of a Mouser catalog faster than trying to use their online web stuff.
The Load Controls device is a power monitor with output signal voltage isolated from the current and voltage inputs, and can be described as a DC voltage proportional to the input power for a steady-state input. Response time is a fractional part of a cycle on a three phase input. I used their device as a means to estimate torque on a machine that did not have a mechanical load torque transducer.
For many current sensors I could not determine whether there was a rectified output or not. Too much superficial and incomplete information on the various websites.
.
