Circuit protection is an area of electronics design that should not be understated. Circuits that do not implement protection features are vulnerable to damage. A good design is not just about how good your circuit is at mitigating damage; it can also be about how well the design protects others.
A new dimension of circuit protection
When it comes to circuit protection the most important goal is to ensure that a design is protected from external sources as well as from itself. The most common source of damage to modern electronics is from static electricity (such as clothes rubbing against skin). This is often mitigated with the use of diodes.
Circuits can also damage themselves if designed incorrectly. One common example is the inductor — a switching relay can produce very large back EMFs. Back EMFs can damage silicon devices. This is why diodes are placed in parallel with coils and inductors; to ensure that the energy released from the sudden collapse of the magnetic field is conducted across the diode instead of the collector or drain of a transistor.
Designers who want to take their circuit protection a step further will consider how their circuit behaves in fault scenarios. A classic example is the inclusion of a fuse in series with the power circuitry. This approach prevents the circuit from drawing dangerous amounts of current should a component or connection short circuit. If taken further, designers can implement complex power management systems including a residual-current device, RCD, which can check to see if there is any current leakage, as this may be an indication of incorrect installation.
One key area that designers often overlook is how their circuit may impact other connected circuits that they themselves have not designed. What if, instead of just being concerned about their own circuit, they considered all other circuits around them?
Becoming a more considerate designer
Many circuits in production are individual and closed. This means they do not physically connect or interact with other devices. However, they are many circuits which do interact with external devices (such as peripherals), and in many cases these peripherals are not designed by the engineer responsible for the original circuit. A good engineer can protect their own circuit, but a great engineer also considers how their circuit could protect other circuits. Expecting an engineer to design circuit protection with external peripherals in mind that they themselves did not design is arguably unfair. How does the engineer know that the connected peripheral follows common protection standards?
Potential protective methods
As seen in previous articles, circuit protection can be more than just basic circuitry and power management ICs. With so many designs incorporating microcontrollers, the processing power can be used to perform a wide range of additional tasks involving circuit protection.
Instantaneous monitoring
Connected peripherals that require a power source (from the host circuit) can be monitored for absolute voltage and current readings. The bus specification (determined by the original design engineer) can have defined voltage and current ranges. If these defined ranges are exceeded, the result will be a power disconnect to prevent damage from short-circuits.
Communication bus
Protection circuitry (such as clamping diodes) are very useful in situations where the expected voltage and current ratings are well understood. Take, for example a host, designed to connect to external thermistors or photoresistors. Connections that may connect to more obscure devices with multiple functions such as sensors, are far harder to protect against.
For example, a digital bus such as I2C, may have a wide range of potential devices, including memory (which draws little current) and motor drivers (which draw large amounts of current). In these situations, simple fuses and current limiters are of little help as a memory chip may be broken with 100 mA, but a motor driver could require amps before a motor begins to turn. In this situation, the host needs to be able to dynamically set current and voltage limits depending on the connected device. Therefore, a communication bus should be implemented (whether it can be superimposed on the already existing digital bus or a dedicated one designed) between the host and the connected peripheral. The bus could then specify a wide range of commands and ping signals, which, in turn, could provide the host with valuable power information. (e.g. current and voltage settings can be sent by the peripheral so the host knows what the current trip point should be set to). Another use of the bus would be a keep-alive signal whereby a properly working peripheral (based around a microcontroller) periodically sends data packets to the host. If the peripheral microcontroller fails due to a short circuit on its power rail, then the host can detect this as the microcontroller will no longer be sending keep-alive data packets.
Artificial intelligence and predictive monitoring
Microcontroller technology has drastically improved in the past decade, with many containing 64-bit ARM cores and large amounts of RAM and ROM. With the rise of IoT and smart technologies, many microcontrollers are being integrated with artificial intelligence (AI) processing modules designed to run neural nets and other machine learning algorithms. While these AI modules are aimed at being used for processing user data, such as speech patterns and sensory behaviour, they can also be potentially used for circuit monitoring. While no current system exists, AI could be used to monitor voltage and current levels to recognise patterns including connection, disconnection, and increased power draw during periodic operation (such as wake up/operate/sleep cycles). Deviations from the normal operation could be used to display warnings of a potential failure or prevent significant power draw from the connected device. This method of circuit protection could be highly beneficial as some forms of circuit damage can take time to occur but present non-critical problems beforehand.
For example, a faulty relay may have sticky contacts, which would present itself as increase current draw (time needed to activate the coil before it switches). An AI system would be trained to recognize typical circuit operation and the increased coil switch on time would be recognized by the AI system as an issue.
Conclusion
Protecting circuits from damage is critical for products to be reliable. Considering all the devices that could potentially be connected to that product is going beyond the call of duty. Products that protect themselves and the devices connected to them can massively boost customer confidence in the product and therefore provide a potential unique selling point.
