I’ve been working with a customer whose product had significant broadband radiated emissions throughout the spectrum, helping them with fault finding and improving their EMC performance. Partial information from their certifying test lab appeared to show a regular harmonic series suggesting noise from a digital clock of some kind. However, near field probing with a spectrum analyser on the bench disproved this theory and identified the brushed DC motor as the cause. It is believed that a combination of the certifying test lab only performing one frequency sweep, the long cycle time of the EUT and the random arcing noise from the DC motor brushes caused a series of regular peaks to appear distributed throughout the spectrum.
In reality, both spectrum analyser measurements and the radiated emissions plot shown below show a wide range of broadband noise once the graph had multiple sweeps through the spectrum to enable it to become fully populated.
Working closely with the customer, we selected an alternate motor from the same manufacturer part with suppression capacitors built in to decouple the brush noise at the source; always the most efficient way of solving an EMC problem is to knock it on the head where is is being generated. Putting the EUT in the anechoic chamber showed that this provided a significant improvement, especially at higher frequencies but there remained an hump at around 150MHz to 300MHz.
We followed up with testing various ferrite cores around the DC motor power leads to further reduce the noise being conducted down the lead, leading to the selection of a small ferrite core with two turns around the motor DC supply leads.
Below can be seen the scan data showing before any modifications (blue) and after changing the motor and adding the ferrite core (green). Testing also revealed what appeared to be a satisfied customer 😉