In part 1 of this mini-series, we noted that taking current measurements may necessitate some form of electrical isolation. In addition to safety, isolation may also be required to eliminate ground loop problems and semiconductor breakdown.
As part of our discussions, we introduced the concept of opto-isolation, but other methods of isolation are possible. If you are dealing with AC, there is always the option of a current transformer (CT). Quite a few EE graduates are unaware that these exist. They are analogous to voltage transformers in that the output current is a ratio of the input current. Because of the magnetics involved, the frequency range is quite limited for a given CT, but this is not a problem in most cases. Typically, the current is converted to a voltage by passing it through a resistor (sometimes called a burden resistor) in the same way as the current sense amp, but with the advantage that the one side of the secondary can be referred to ground, so a differential conversion is not required.
In the industrial world, there is a whole range of CTs that transform a particular input current to an output current of 5 A. That will likely require a higher-wattage load resistor because of the power dissipated, but there are other options, like those made by Bicron and CR Magnetics. I have used products from both companies, and they work as advertised. Some CT specs will limit the voltage developed across the load resistor to a maximum value -- make sure you observe such a restriction.
Some CTs are arranged as clamps that let you clip the CT around a wire without disconnection. All the ones I have seen like this are part of a DVM or scope setup, and they tend to be temporary. For permanent fixtures, you will need to feed a flying lead through the center of the toroid. This can be inconvenient and often is untidy. Also, if the secondary of a CT is left open, the voltage can climb to shocking levels. In electrical panels, standards organizations like UL or CSA can require a special shorting terminal to connect the CT ends if the load is removed. The standards can also require that the one side of the CT be earthed.
This brings us to Hall effect transducers, which also provide isolation along with the versatility to measure both AC and DC. I am not going to get into the theory of operation of a Hall effect transducer here, except to say that the relationship between the output and input current is a function of a physical gap, normally in a ferrite toroid. As a result, the Hall effect transducer will typically have a relatively low accuracy and a susceptibility to changes from temperature swings.
Honeywell makes an enormous number of Hall effect transducers, many of them PCB mounting. I found it very difficult to find a summary on the Honeywell website, so I worked from Digi-Key. It appears that, as with CTs, you have to feed a flying lead through most of the Honeywell sensors.
LEM is another participant in the current sensor market, principally working with Hall effect transducers. Some are like CTs where the current lead has to be fitted through the core. LEM used to sell a packaged Hall effect transducer with the specification that the board trace ran immediately under the IC package. This approach doesn't seem to have been too popular, because it no longer seems to be available. LEM has one device (or more maybe) where the Hall effect sensor is embedded in the IC package, and the current tracks run up to pins on the package. The output is an analog signal biased around the supply half rail. See the HMS family for more details. LEM is another company whose website is difficult to navigate (at least to me) when you are trying to find specific parts. Thus, as before, I resorted to Digi-Key's selection.
To Page 2 >