A recent Brad Albing post on this site covered the topic of signal isolation. It reminded me that I had done a blog in a similar vein at Microcontroller Central. I thought I might revisit that blog and update it.
There are two main reasons to use galvanic isolation, especially in analog applications. The first is the phenomenon known as a ground loop, where one circuit's ground reference or voltage potential is different from that of a second circuit. This occurs for various reasons — the effectiveness of connecting to zero potential, external electrical and magnetic fields, leakage currents, etc. I have seen tens of volts difference between grounds only 100 yards apart. The resulting current can be extremely high when the only limiting factor is the wire you are using to short the two. Even a small current can corrupt the analog voltages you are measuring by adding offsets, which may even vary dynamically.
The second reason for using galvanic isolation is the need for safety and/or the need to remove a high common mode signal. If you are measuring a current on the high side of a high-voltage motor, you might be taking a measurement of several hundred volts.
In many cases, you also may need to provide an isolated power supply to drive the isolated electronics, but I am not going to focus on that. You may want to look at Analog Devices (ADI) isoPower devices as a starting point, but there are many other options.
My first encounter with an isolated analog system has stayed with me a long time. It was a Multibus (an Intel single-board computer standard) board made by Burr Brown. A capacitor was connected to the common pins of a two-pole changeover reed-relay. The normally open (NO) pins of the relay were connected to the differential input signal. The NC pins were connected to the differential input pins of the ADC. The logic would switch the relay to the NO position, allowing the capacitor to charge. The relay was de-energized, and the voltage on the capacitor was presented to the ADC. This would be quite effective from an isolation perspective, but it could not handle even moderate frequencies. There are better solutions today.
A fairly old approach is magnetics. If you had an AC signal, you could simply use a transformer, but DC signals required a different approach. ADI always has had a strong presence in this field. It has complete modules that include the transformers like the AD208 and the AD202.
These parts give excellent performance but are not cheap, given the complexity of their construction. TI also makes some capacitor coupled isolation devices, like the ISO124. LEM seems to have modules in this field — the AV100 series.
In the next part of this series, we will continue looking at the use of magnetic coupling to transfer signals.