Radiocommunications Agency  EMC Awareness

Circuit design for RF stability

Introduction

Radio emissions need not always come from intentional RF oscillators, whether they be transmitters, receivers or digital products. Although conventional analogue circuits – like audio or instrumentation amplifiers – have no intentional way of generating RF, they can still do so if improperly designed. And often, when this happens, it isn’t realised by the product designer, because it doesn’t affect the circuit operation, and it isn’t discovered in EMC testing, because many standards still allow such circuits to avoid emissions checks.

Key issues

Circuit design

Any amplifier can oscillate if it has positive feedback from input to output. Criteria for the stability of feedback amplifiers were developed many decades ago. Roughly speaking, if a feedback circuit’s phase shift is such as to give positive (in phase, as opposed to anti-phase) feedback at a frequency at which the amplifier’s gain is still greater than unity, then oscillation will occur. Many operational amplifiers have gain bandwidths of tens or hundreds of MHz; some, such as video amplifiers, have wider bandwidths still.

Feedback paths can be intentional or stray: stray paths include capacitance from output to input, and high impedance power terminals due to inadequate decoupling. Only small values of capacitance and inductance are needed to give uncontrolled phase shifts in the MHz region, and these can be layout-related.

The circuit need not be oscillating at a high level to cause unacceptable emissions: only a few mV amplitude may be sufficient in some circumstances, and this may have no discernible effect on the circuit’s desired LF operation.

Mitigation

The designer should be aware of the potential problem and should ensure that circuits and PCBs are designed to avoid it. This means, particularly, care in layout and decoupling of wideband amplifiers. The applications notes for such devices often give detailed guidance for ground plane layout and power decoupling, and this should be taken seriously. Further to this, some testing of RF emissions should be performed on analogue circuits, even if it isn’t needed for compliance with standards, to prove that low-level RF oscillation is absent. Another hint can be gained from RF immunity testing: if a circuit is especially susceptible at or near a particular frequency, this may be because the amplifier is only marginally stable at this frequency, and even if that particular example is not emitting, another unit with a slightly different build may do so.