When Noise Crosses a Common Mode Choke Through Its Windings

A common mode choke is placed on a power input to block common mode noise, and at low frequency it does exactly that: the specified inductance is large and the impedance it presents to common mode current matches the datasheet curve. At higher frequency the noise reappears downstream as though the component were not there at all. The path it took is usually not through the magnetic circuit but between the windings, through capacitance that the ideal model omits.

Every turn of wire has capacitance to the turns beside it and to the turns in the other layer. The input terminal and the output terminal sit close together on the same former, and the windings of the two lines are separated only by insulation. At low frequency those capacitances are negligible. As frequency rises their impedance falls, and at some point the noise finds a lower impedance path across the component than through it.

The Choke Is a Distributed Network

A real choke contains winding resistance, leakage inductance, turn-to-turn and layer-to-layer capacitance, and core loss. The two coupled inductors of the ideal model are only the low frequency part of that network. Common mode current sees the magnetising inductance and is attenuated; differential mode current largely cancels in the core but still encounters leakage inductance and the parasitic elements.

The practical consequence is that adding turns is not a reliable way to improve high frequency performance. More turns increase the inductance, which improves the low frequency impedance, and they also increase the capacitance between the windings, which lowers the frequency at which the parasitic path becomes dominant. The two effects move in opposite directions, and beyond a certain point the extra inductance is offset by the degraded high frequency behaviour.

Self Resonance Sets the Useful Band

The impedance of the component rises with frequency until the capacitance across the windings resonates with the inductance, after which the impedance falls and the part behaves capacitively. That self resonance frequency is the upper limit of the band in which the choke can be expected to attenuate anything, and the impedance peak at resonance is finite. Above the resonance, the component can even amplify the coupling it was installed to prevent.

This is why a selector should compare parts at the frequency of the problem rather than at a nominal measurement frequency, and why a part with a lower inductance can outperform a higher inductance part in the band that matters. The impedance curve across frequency, including the resonance, is the specification that answers the question; the inductance value alone describes only the low frequency end.

The Most Dangerous Path Is Input to Output

The most harmful parasitic is the one that connects the input terminal to the output terminal, because it allows noise to cross the component without entering the magnetic circuit at all. It is formed by the capacitance between the start of one winding and the end of the other, and it grows when the winding start and finish are placed close together on the former, when the layers overlap over a large area, or when the two windings are wound together rather than in separated sections.

Inter-winding capacitance cannot be eliminated, but it can be reduced by construction. Sectioned bobbins, split windings and physical separation between the input and output ends all raise the impedance of the path across the component. For a given part, the relevant specification is the impedance measured from input to output with the windings otherwise unconnected, and it is a figure worth requesting when the application needs attenuation at high frequency.

Board Level Paths That Bypass the Part

The component can also be bypassed outside itself. If the trace before the filter and the trace after it run parallel for even a short distance, the capacitance between them provides a path around the choke. The same happens when copper pours on adjacent layers overlap across the isolation boundary, and when a connector or a downstream cable runs close to the unfiltered side. Common mode noise does not recognise the drawing on the schematic; it follows the lowest impedance closed path available.

The layout response is to treat the filter as a boundary. Keep the dirty side and the clean side physically separate, avoid any copper that crosses the boundary other than the intended connection, and keep test points and protection devices out of the gap. The return path to the noise source should be controlled rather than incidental, since the current has to return somewhere and the route it takes is the route that couples. These principles are the same as those set out in EMI suppression principles, and the reference plane arrangement that supports them is discussed in ground current and harmonic distortion.

Practical Points

Filter placement decisions followed by a measurement of the cleaned side, rather than of the component in isolation, are what confirm that a choke is doing its job. Where the assembly is coated, verify that the coating does not introduce a leakage path across the isolation gap, as described in conformal coating protection.

Process Control and Verification

Reviewing the design before the data is released is far cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end. Where a value sits close to a process limit, the drawing should say so, since the shop can then open the process window rather than working to a nominal figure that carries no tolerance.

The measurements that matter are the repeatable ones: conductor width and spacing, annular ring, finished hole size, plating thickness and surface finish are all verifiable on a coupon that travels with the panel. Documenting the assumption is part of the design work, and a short note on the drawing prevents a question that would otherwise arrive a day later and cost a day of schedule.

Where a value sits close to a process limit, the drawing should say so, since the shop can then open the process window rather than working to a nominal figure that carries no tolerance. Running a first article through the same checks as the production panel confirms that the two agree, and that comparison is the cheapest form of process control available at prototype stage.

Documenting the assumption is part of the design work, and a short note on the drawing prevents a question that would otherwise arrive a day later and cost a day of schedule. A stack-up that is drawn rather than described removes most of the ambiguity from a quotation, and it lets the fabricator price the board against the dielectric and copper weights that will actually be used.

FAQ

Why does a larger inductance not always filter better? Because the extra turns also increase the capacitance between the windings, which lowers the self resonance frequency. Above that frequency the component no longer attenuates, so a higher inductance part can be worse in the band of interest.

Can the windings be separated to reduce the coupling? Yes. Sectioned winding and physical separation between the input and output ends reduce inter-winding capacitance at some cost in leakage inductance and size, and the trade should be evaluated at the frequency that matters.

How is the choke verified after assembly? Measure the noise on the clean side with the product operating normally, then repeat with a near field probe to see whether coupling is occurring around the component rather than through it.