**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 50Ω 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):

(2.1)

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,

and

(2.2)

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 (50Ω is usual value). Engineers with years of experience in EMC admit these attenuation curves, generally prepared from data taken in a 50Ω test setup, to be of extremely limited value. In spite of the low benefit of this information, manufactures publish 50Ω data, because of the easy measurement procedures and equipment availability, since connectors, test cables and instrumentation characteristic impedance are 50Ω. Attenuation curves using 50Ω 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 0.1Ω and 100Ω terminations on the line and load side, instead of 50Ω termination, measuring the filter insertion loss. Afterwards the measurements are repeated changing the termination impedances 100Ω and 0.1Ω 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 50Ω 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.

(2.3)

where ZS- source impedance,

ZL- load impedance,

ZT- filters impedance,

and

(2.4)

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:

(2.5)

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):

(2.6)

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