Radiocommunications Agency  EMC Awareness

Preventing unintentional RF demodulation

What this technique is used for

Unintentional demodulation is a major cause of RF immunity problems in both analogue and digital circuits, and this technique helps reduce its effect.

Radio, TV and radar receivers must demodulate the signals to make them useful to people. Demodulation involves stripping an RF signal of its carrier frequency so that its modulation envelope (the bit we are interested in) becomes ‘baseband’ and is then useable as data, sound, picture, range, whatever.

However, if a circuit is not meant to be a radio, TV or radar receiver – any RF signals that it demodulates will be unwanted electrical noises that are hard to distinguish from the wanted signals.

When an unmodulated RF signal gets demodulated it creates a d.c. offset, whilst a modulated signal creates both a d.c. offset and the modulation envelope. When the level of this demodulation noise gets too high, the circuit’s performance will be noticeably degraded and we say that it is suffering from interference and doesn’t have enough immunity.

Unfortunately, all non-linear circuits cause demodulation, including diodes and rectifiers, thermistors, transistors and all types of ICs. Also, metal oxides and salts can behave as semiconductors, which means that poor electrical joints can behave as diodes and cause demodulation too.

This unintentional demodulation is sometimes called “rectification” or “audio rectification” to distinguish it from the intentional demodulation in a receiver circuit.

The trouble is that there are a lot of intentional RF signals around – radio and TV transmitters, mobile phones, radars, and other sources such as industrial RF processing, most of them emitting modulated signals. Electronic circuits can be subjected to these signals at such high levels that unintentional demodulation is a real issue.

How this technique is used

This technique is used by circuit designers to improve the RF immunity of their designs to reduce the demands on costly and space-consuming filtering and shielding techniques.

Key issues in employing this technique

Metal joints

Metal-to-metal joints should be designed and assembled so as to prevent oxidation and corrosion between their contacts. Often this just means that suitable metals and surface coating/plating should be chosen and then pressed together with sufficient force over the life of the equipment (or soldered). But where metal joints are exposed to water or other liquids some additional physical protection might be required.

The metal-to-metal joints to be considered should include all of the various metal parts involved in an equipment’s construction, and where the RF fields are intense this may need to be extended to support brackets and other nearby metal parts. Also refer to the section on intermodulation for more on improving metal joints.

Semiconductors

Some devices are more prone to demodulation than others, for instance bipolar opamps tend to demodulate about 20dB worse than bi-FET, JFET or CMOS types), but semiconductor manufacturers do not provide data that makes it possible to directly compare devices for this aspect of their performance.

Digital ICs usually have greater immunity than analogue types to low-level demodulation noise because of their digital signal thresholds. But when the demodulated noise exceeds the signal threshold digital circuits can cause software to malfunction or ‘crash’ with potentially serious consequences. Even below the threshold level, interference can create timing errors that cause failure in marginal designs.

It may be best to test the devices you are thinking of using to see how immune they are to RF. Standard test methods are being developed for this purpose, but it may be best to set-up the test to correspond roughly to the way the device will be used in the final product.

Circuit design

The effects of d.c. offsets can sometimes be reduced by using a.c. coupled circuits, up to the level at which the shift in operating point causes distortion of the wanted signal. Simple passive filtering circuits (with RC networks) can be used to reduce the amplitude of RF interference at sensitive rectifying nodes such as op-amp inputs, outputs and power supply pins.

The modulation envelope is more of a problem, as this can lie in the same frequency band as the wanted signal. In some types of instrumentation it is possible to use a.c. energised sensors and a phase-sensitive detector (sometimes called synchronous rectifier) to reject a great deal of noise.

Stability

Many feedback circuits are not very stable at RF, and many opamps can amplify or even oscillate at frequencies 10 times higher than their gain-bandwidth product would suggest – increasing the levels of demodulation in later devices. Component tolerances, different types or lengths of cable, and temperature variations can have a big effect on a feedback circuit’s stability, so it is important to ensure that circuits are designed to remain stable despite all these variations.

It is also useful to make feedback circuits as wideband and fast as possible, whilst still remaining stable. This increases the range of frequencies at which they behave linearly, and circuits demodulate the least when used in their linear frequency range.