How to be a good analog engineer (3): Digital isolation and analog isolation

Industrial circuit design engineers use isolation techniques to address security issues, regulatory oversight, and ground plane issues. If isolation is done in your circuit, you can exchange information and power between two points without actual current flow. There are two major benefits to isolation. First, it protects people and equipment from potentially dangerous surge currents and voltages. Second, it prevents accidental ground loops that interfere with signals from data links and other interconnects. The e-family network has detailed and explained based on several common isolation techniques.

Introduction to digital isolation technology

Designers of industrial, medical, and other isolation systems have had limited means of achieving safe isolation for many years, and the only reasonable choice is optocouplers. Today, digital isolators have advantages in performance, size, cost, efficiency, and integration. Understanding the characteristics of the three key elements of a digital isolator and their interrelationships is important for the proper selection of digital isolators. These three elements are: insulation, structure and data transmission methods.

Designers introduced isolation to meet safety regulations or reduce ground loop noise. Galvanic isolation ensures that data transmission is not through electrical connections or leak paths, thus avoiding safety risks. However, isolation introduces limitations in latency, power consumption, cost, and size. The goal of digital isolators is to meet safety requirements while minimizing adverse effects.

Traditional isolators—optocouplers can have very large adverse effects, high power consumption, and data rates below 1 Mbps. Although there are higher efficiency and higher speed optocouplers, the cost is higher.

Digital isolators were introduced more than 10 years ago to reduce the adverse effects associated with optocouplers. Digital isolators use CMOS-based circuits to provide significant cost and power savings while dramatically increasing data rates. Digital isolators are defined by the above elements. Insulating materials determine their inherent isolation capabilities and the materials selected must meet safety standards. The choice of structure and data transmission methods should aim to overcome the above-mentioned adverse effects. All three elements must work together to balance the design goals, but one goal must be achieved without compromise, that is, compliance with safety regulations.

Insulation Materials

Digital isolators are fabricated in a wafer CMOS process and are limited to common wafer materials. Non-standard materials can complicate production, resulting in poor manufacturability and increased costs. Commonly used insulating materials include polymers (such as polyimide PI, which can be spin-coated into a film) and silicon dioxide (SiO2). Both have well known insulating properties and have been used for many years in standard semiconductor processes. Polymers are the basis of many optocouplers and have a long history as high voltage insulators.

Safety standards typically specify a 1-minute withstand voltage rating (typically 2.5 kV rms to 5 kV rms) and an operating voltage (typically 125 V rms to 400 V rms). Some standards also specify shorter durations, higher voltages (such as 10 kV peaks for 50 μs) as part of the reinforced insulation certification. Polymer/polyimide based isolators improve optimum isolation characteristics as shown in Table 1.

Polyimide-based digital isolators are similar to optocouplers and last longer at typical operating voltages. SiO2-based isolators are relatively weak against surges and cannot be used in medical and other applications.

The inherent stress of various films is also different. The stress of the polyimide film is lower than that of the SiO2 film, and the thickness can be increased as needed. The thickness of the SiO2 film is limited, so the isolation capability is also limited; when it exceeds 15 μm, the stress may cause the wafer to crack during processing or delamination during use. Polyimide-based digital isolators can use isolation layers up to 26 μm thick.

Isolator structure

Digital isolators use a transformer or capacitor to magnetically or capacitively couple data to the other end of the isolation barrier, which uses LED light.

The transformers feature a differential connection that provides excellent common-mode transient immunity up to 100 kV/μs (optocouplers are typically around 15 kV/μs). The dependence of the magnetic coupling on the distance between the transformer coils is also weaker than the dependence of the capacitive coupling on the distance between the plates. Therefore, the insulating layer between the transformer transformer coils can be thicker, thereby achieving higher isolation capability. In combination with the low stress characteristics of polyimide films, transformers using polyimide are easier to achieve advanced isolation than capacitors using SiO2.

Capacitors are single-ended connections that are more susceptible to common mode transients. Although it can be compensated with a differential capacitor pair, this will increase the size and increase the cost.

One of the advantages of a capacitor is that it uses a low current to create a coupled electric field. When the data rate is high (above 25 Mbps), this advantage is quite obvious.

Data transmission method

The optocoupler uses the light from the LED to transmit data to the other end of the isolation barrier: when the LED is lit, it indicates a logic high level, and when it is off, it indicates a logic low level. Optocouplers need to consume power when the LED is lit; optocouplers are not a good choice for applications that focus on power consumption. Most optocouplers leave the signal conditioning at the input and/or output to the designer, which is not necessarily a very simple task.

Digital isolators use more advanced circuitry to encode and decode data, support faster data transfer speeds, and handle complex bidirectional interfaces such as USB and I2C.

One method is to encode the rising and falling edges as a double pulse or a single pulse to drive the transformer (Figure 2). These pulses are decoded on the secondary side as rising or falling edges. This method consumes 10 to 100 times less power than optocouplers because unlike optocouplers, the power supply does not need to be continuously supplied to the device. A refresh circuit can be included in the device to periodically update the DC level.

Another approach is to use an RF modulated signal that is used in much the same way that an optocoupler uses light, which would cause continuous RF transmission. This method consumes more power than the pulse method because the logic high signal requires continuous power consumption.

Differential techniques can also be used to provide common mode rejection, but these techniques are best used with differential components such as transformers.