Con­ven­tional char­ac­ter­i­za­tion of fil­ters

Elec­tro­mag­netic com­pat­i­bil­ity (EMC) is manda­tory for all elec­tronic devices to access the Euro­pean Union and other mar­kets. Elec­tronic devices cre­ate elec­tro­mag­netic envi­ron­ment around them self and expose other equip­ment. Espe­cially indus­trial and power elec­tronic devices are pow­er­ful dis­tur­bance sources and their per­ilous gains due to pos­si­bil­ity of portable move­ment and con­nec­tion to almost all power sup­ply grids. To com­ply with one of the emis­sion types– con­ducted emis­sions, power line fil­ters are used in nearly all elec­tronic devices. Power line fil­ters are selected to atten­u­ate con­ducted emis­sion dis­tur­bances in defined fre­quency range, which depends on the dis­tur­bance source. Fil­ter atten­u­a­tion in fre­quency range is defined as inser­tion loss. Fil­ters inser­tion loss is depen­dent value. It depends on source and load imped­ances, thus fil­ters per­for­mance in real elec­tronic sys­tem can­not be pre­dicted, with­out knowl­edge of these para­me­ters. Usu­ally inser­tion loss of fil­ter is mea­sured using 50 ter­mi­na­tion in load and in source. More­over, an inter­na­tional stan­dard CISPR 17 is pub­lished that defines other inser­tion loss mea­sure­ment tech­niques called “Approx­i­mate worst case”. Despite this method is used for char­ac­ter­i­za­tion of fil­ters widely, there is a neces­sity to develop, improve and use char­ac­ter­i­za­tion method that is inde­pen­dent of noise source and load imped­ances.

EMI fil­ters are typ­i­cally char­ac­ter­ized by their inser­tion loss, which is usu­ally stated in deci­bels (dB). Fil­ter is typ­i­cally inserted between the source of the dis­tur­bances and load, in order to pre­vent the unwanted dis­tur­bance sig­nals to affect the per­for­mance of load. The sit­u­a­tion is shown in Fig. 2.1.

Fig. 2.1. Typ­i­cal appli­ca­tion of power line fil­ter

Load volt­age with fil­ter are denoted by V(L,w), but load volt­age with­out fil­ter are denoted by V(L,wo). There­fore, the inser­tion loss is defined by equa­tions (2.1) and (2.2):


where ILdB- inser­tion loss in dB,
PLwo- power on load with­out fil­ter,
PLw- power on load with fil­ter,
PLw- power on load with fil­ter,
VLwo- load volt­age with­out fil­ter,
VLw- load volt­age with fil­ter,
RL– load resis­tance,


Inser­tion loss reduces the volt­age due to the inser­tion of the fil­ter, at the fre­quency of inter­est. Fil­ters spec­i­fi­ca­tion are often given assum­ing that the source and load imped­ances are equal to some spec­i­fied value (50 is usual value). Engi­neers with years of expe­ri­ence in EMC admit these atten­u­a­tion curves, gen­er­ally pre­pared from data taken in a 50 test setup, to be of extremely lim­ited value. In spite of the low ben­e­fit of this infor­ma­tion, man­u­fac­tures pub­lish 50 data, because of the easy mea­sure­ment pro­ce­dures and equip­ment avail­abil­ity, since con­nec­tors, test cables and instru­men­ta­tion char­ac­ter­is­tic imped­ance are 50. Atten­u­a­tion curves using 50 imped­ance are often crit­i­cized in many books and tech­ni­cal papers as well as in inser­tion loss mea­sure­ment stan­dards such as Mil Std 220 and CISPR 17. CISPR 17 pro­poses an alter­na­tive mea­sur­ing method the so called “Approx­i­mate Worst Case Method”. This test method uses 0.1 and 100 ter­mi­na­tions on the line and load side, instead of 50 ter­mi­na­tion, mea­sur­ing the fil­ter inser­tion loss. After­wards the mea­sure­ments are repeated chang­ing the ter­mi­na­tion imped­ances 100 and 0.1 ter­mi­na­tions on the line and load side. Although this test method is not the same as mea­sur­ing a fil­ter in a real equip­ment instal­la­tion, the nor­mal­ized results can be used with rel­a­tive accu­racy to pre­dict the per­for­mance of the fil­ter in a real sit­u­a­tion. Another advan­tage of CISPR 17 — mea­sure­ment method is well defined, that leads to accu­rate and repeat­able results. The power line fil­ter indus­try must pub­lish data on its prod­ucts using rec­og­nized, stan­dard­ized and accepted test meth­ods. If, as gen­er­ally accepted, the 50 method can­not be used to pre­dict the per­for­mance of a fil­ter in real equip­ment, the CISPR 17 “Approx­i­mate Worst Case” method is the only such stan­dard­ized test to meet this require­ment.

To deter­mi­nate the actual inser­tion loss in dB, if the source and load imped­ances are known, equa­tion (2.3) or (2.4) can be used, depend­ing on the fil­ters char­ac­ter­is­tic trans­fer imped­ance. Equa­tion (2.3) is valid in case, if the fil­ter is used as shunt or in par­al­lel with load and source.


where ZS- source imped­ance,
ZL- load imped­ance,
ZT- fil­ters imped­ance,

The imped­ance ZT would be equal to the ratio of the volt­age across the open cir­cuited out­put of the fil­ter, to the cur­rent into the fil­ter. Using volt­age divi­sion and alge­bra, inser­tion loss can be obtained as fol­lows:


where VS- source volt­age.

As an exam­ple for shunt fil­ter the capac­i­tor can be men­tioned. Thus, capac­i­tor imped­ance should be mod­eled as RLC cir­cuit rep­re­sent­ing real capac­i­tor in applic­a­ble fre­quency range.

Equa­tion (2.4) is valid in case if the fil­ter is used in series with load source. Fil­ters inser­tion loss can eas­ily be obtained with equa­tion (2.6):


Fil­ters used in series for power elec­tronic are fer­rite chokes and beads.
eXTReMe Tracker