The most common core shape used in a common mode inductor is
the toroid. It has two big advantages compared to other shapes. First, if you
think of it as a winding on one half of a toroid opposing a winding on another
half of a toroid you have two very symmetrical shapes which helps prevent core
saturation. Secondly, a toroid has no mating surface compared to other core
shapes so it will have more inductance and is usually 35% higher in inductance
compared to a similar sized two piece core shape. The two windings on the
toroid need to be as symmetrical as possible since leakage inductance will be
created by the current flow. Leakage inductance can start to saturate the core
which will cause the common mode inductance to decrease. An unequal number of
turns on the two windings or an unsymmetrical winding shape can increase
leakage inductance resulting in reduced inductance leading to reduced filtering
of the noise. For more details see the theory of common mode inductors.
Typically common mode inductors handle currents ranging from
tens of mA to about 20A. This means wire gauges from #38 to #12 are commonly
used. Currents much above 20A are usually not wound on a toroid since the core
can break due to winding stress associated with heavy gauge wire. Other core
shapes and unique windings are usually used above 20A. Also multiple inductors
in parallel could be used. High permeability ferrite materials are very
susceptible to winding stress and inductances can be down to 50% of the
expected inductance due to the winding stress on the core. Usually heating the
wound cores slowly relaxes the stress and brings the inductance back up. I
emphasize that the heating should be ramped up and down slowly since ferrite is
a ceramic and can crack due to rapid temperature changes.
One of the more surprising aspects of toroidal common mode
inductors, or toroids in general, is that they are wound automatically, not by hand on toroidal winding machines.