Arcs and sparks (transients) occur at many electromechanical contacts, such as switches, relays, contactors and the commutators of brushed electric motors.

When current through an inductance is interrupted, a large transient voltage is generated, governed by V = – L·di/dt.  Theoretically if di/dt is infinite then the voltage is infinite too; in practice it is limited by stray capacitance if no other measures are taken. Switching can either be via an electromechanical contact or a semiconductor, and the latter can easily suffer avalanche breakdown if the transient is unsuppressed. RF interference is generated in both cases at frequencies over the entire RF spectrum determined by stray circuit resonances and is usually coupled by the wiring between switch and load.

Such emissions can be dealt with by shielding and filtering, but these are expensive techniques. If the RF energy can be reduced at source, the cost of filtering and shielding required can be reduced or even eliminated. This can have the side-benefit of increasing the operational life of the electromechanical contacts.

Contact suppression

Arcing and sparking at electromechanical contacts is caused by the ‘flyback’ voltage generated by the inductance in the circuit when the electromechanical contact tries to interrupt the circuit’s current abruptly. The flyback voltage is so high that it causes the air in the contact gap to break down and an electrical discharge (an arc or a spark) to occur.

An opening electro-mechanical contact which interrupts a flow of current – typically a switch or relay – will initiate an arc across the contact gap. The arc will continue until the available current is not enough to sustain a high voltage across the gap. As the gap increases while the contacts separate, its withstand voltage increases, with two regions evident, the earlier one characterised by arc discharge and the later one by glow discharge. The stray capacitance and inductance associated with the contacts and their circuit will in practice cause a repetitive discharge until their energy is exhausted, and this is responsible for considerable broadband interference. A closure can also cause interference because of contact bounce.

All transient suppression techniques use circuit components to slow the rate of change of the current and thereby reduce the magnitude of the flyback voltage. These are sometimes called ‘snubbers’, and the technique called ‘snubbing’.

The conventional suppression circuit is an RC network connected directly across the contacts. The capacitor is sized to limit the rate-of-rise of voltage across the gap to below that which initiates an arc, typically 1V/µs for most contacts. The resistor limits the capacitor discharge current on contact closure; its value is a compromise between maximum rated contact current and limiting the effectiveness of the capacitor. It should have a value substantially lower than the impedance of the load, otherwise a large proportion of the transient voltage will be dropped across it rather than caught by the capacitor. A diode in parallel with the resistor can be added in DC circuits if this compromise cannot be met.

The RC snubber circuit is most typical, but other circuits use diode, Zener or varistor clamps as shown. Their disadvantage is that they may affect the turn-off time of the circuit, since they divert the inductive transient current on turn-off and allow it to flow for a short time afterwards. In all cases the suppression components must be mounted immediately next to the load terminals, otherwise a radiating current loop is formed by the intervening wiring.  Protection of a remote drive semiconductor must be considered separately.

Where is the inductance?

It is usually the load inductance in the circuit that ‘flies back’ when the current is interrupted and so creates the transient at the opening electrical contacts. Identifying where the inductance is helps to apply the mitigation technique. Inductive loads which tend to cause arcing and sparking include relay and contactor coils, and the coils of electric bells. It is generally better for EMC to suppress an inductive load at the load itself, so that the current path for the high-frequency components of the flyback waveform covers the smallest possible area and so radiates least.

Mains-powered applications have a considerable amount of stored energy in the inductance of the mains supply cables themselves, and this can also be the cause of transients coupled along the supply wiring - resulting in the EFT/Burst, which in turn requires immunity control for potentially susceptible equipment.

Motor noise

DC brushed motor noise is particularly aggressive, since it consists of impulsive and hence wideband transients repeated at a rate determined by the commutation speed – in other words, several hundred to several thousand times a second. The spectral composition of this noise may extend up to several hundred MHz. This appears both as differential mode noise across the terminals and common mode with respect to the housing, coupled through stray capacitance.

The best method of suppression is to prevent the motor from generating impulsive voltages across the commutator segments. This can only be achieved by the motor manufacturer, by incorporating varistor or RC components between each commutator segment, but ensures that the motor is quiet without further suppression being necessary. Otherwise, RC snubbing components across the terminals (for differential mode) and capacitors from the terminals to the case, which is then connected to local earth (for common mode) are required; if the local earth is not available, then a common-mode choke at the motor terminals may be needed.

Other motor types, particularly induction motors, don't suffer from this type of noise generation process. But any motor which is driven from a pulse-width-modulated variable speed drive has to be regarded as a potential source of RF emissions, due to the harmonics of the drive frequency, which is equivalent to a switch-mode power supply with the switching frequency brought out to the motor.