The isolator circuit has gone CMOS. Integrated designs circumvent the bandwidth limitations and other issues of the venerable old opto-isolator.
by Hamilton Carter – Senior Editor
As a kid reading articles in Joe Desposito’s, (scroll down the linked page, he’s there!), Popular Electronics, I often wondered what a circuit isolator was. I was especially curious about opto-isolators. I’d built an optical communications project with an IR diode and a photo-transistor. As far as I could tell an opto-isolator was the essential guts of my project built onto a chip. Because opto-isolators just met the affordability cutoff of my two dollar-a-week allowance, I picked one up at the local parts store and experimented with it. Sure enough, it could serve as an interface between the transmitter and receiver of my little project. The little chip still baffled me though, whereas my original project could transmit across the room, the opto-isolator did the same thing with a somewhat underwhelming range—across the plastic IC package within which it was embedded.
As a ‘grown-up’ engineer I began to understand the appeal of the small devices. Their real intent was to protect, (via isolation… go figure), the more delicate and genteel portions of a system. The digital controller of a motor, for example, could be protected from the large current and voltage spikes that could exist in the motor itself. Maxim Integrated’s, executive business manager, Tony Partow offered this concise definition of the devices and their use:
“Isolators are used in many electrical and electronic systems, mainly for providing safety for system noise management and voltage translation between multiple voltage domains. Maxim Integrated sees this as a rapidly growing market, with applications encompassing many kinds of end equipment including medical systems, factory automation equipment, and electrical grid distribution.”
Isolators are cool; the trusty old opto-isolator has issues though. Silicon Lab’s Ashish Gokhale mentioned pointed out these shortcomings:
Optocouplers are slow. In today’s world of Big Data, this is a serious limitation. Designers want to push more data through the pipeline, and they need more bandwidth, not less.
Optos consume a lot of power. In our age of “green energy,” developers require higher energy efficiency. The high power consumption of optos has created an additional obstacle for designers.
Optos are unstable over operating conditions, especially at high temperatures and with age. For products that served the industrial, automotive or medical markets, this has been a serious limitation.
Not to worry though, the opto-isolator is being replaced design-by-design by integrated CMOS isolators. A variety of enabling technologies are offered by companies such as Maxim Integrated, Silicon Labs and Analog Devices. A recent article by Analog Device’s Dara O’Sullivan and Maurice Moroney nicely highlights the form and function of their particular design: the magnetic isolator circuit.
The Trouble with Opto-Isolators
Opto-isolators present a significant amount of parasitic capacitance at their inputs. For systems that are laid-back and not particularly chatty, this capacitance isn’t really an issue. For controllers that want to frequently update the system however, the capacitance becomes a bit of a problem. The charging of a capacitor circuit is modeled by
where V is the voltage across the capacitor and RC Is the time constant of the circuit. As the time constant becomes larger, the capacitor takes longer to charge as shown below.
The above graph shows the performance issue in the time domain. If you’re trying to drive a square wave through a device with large parasitic capacitance, the signal will have an ever slower rise time as the capacitance increases. This issue limits the frequency of digital signals that can be transmitted through the device.
The (relatively) Speedy Magnetic Transformer Isolators
According to O’Sullivan and Moroney, we can now place a full-fledged coreless transformer on a CMOS chip! If that wasn’t cool enough, the devices provide a solution to the opto-isolator’s data speed issues. Rather than sending signals via electromagnetic radiation, (light), they’re sent by magnetism alone. Two coils are placed in proximity to one another on either side of an intervening isolation material as shown below, (here’s a better look).
The transmitter coil utilizes Faraday’s law of induction to well… induce its signal in the neighboring, but electrically isolated, receiver coil. The all CMOS construction leads to lower parasitic capacitances, allowing faster signal rise times and consequently faster data rates.
The innocuous little isolator circuit has come a long way in a few years. So has our ability to pace ever more interesting structures on CMOS dies.
NOTE: If you enjoyed your undergraduate control classes, you might prefer to think about capacitances in the frequency domain. In this domain, the value of the impedance of a capacitor, is specified as
This might mislead you a bit as it did me earlier today. How could higher frequencies, or capacitances for that matter, be an issue? Increase either of these parameters and the impedance reduces presenting less… impedance, to high frequency signals. The crux of the matter lies in the following diagram that shows the location of the parasitic capacitance.
The less impedance the parasitic capacitance offers, (via an increase in the driving frequency), the more the driving signal is shunted directly to ground and away from the light emitting diode.