Feature: Power electronics
Figure 3: (a) Output impedance of a DC-DC converter; (b) Input impedance of a three-phase voltage source inverter
Figure 4: Current loop in (a) differential mode (b) common mode
the load impedance is defined by the LISN itself, since the noise source originates within the SMPS. From a differential-mode point of view, the LISN can be
treated as having an impedance of roughly 100Ω (two 50Ω resistances in series). If we now consider filter insertion loss, a more realistic impedance model is therefore a variable source impedance feeding into a 100Ω load. This is quite different from the assumptions commonly used in filter datasheets, which typically specify a 50Ω/50Ω system. For common mode, the source impedance depends heavily
on how the product is grounded or earthed. Is the product floating with respect to ground, or is it bonded to chassis? If bonded, how long is the bonding strap or connection? All of these factors directly affect the common-mode source impedance, which like all impedances is also frequency dependent. On the load side, however, the EMC test setup is much more
clearly defined. As shown in Figure 4, the common-mode load impedance in a typical LISN-based setup can be simplified to 25Ω (two 50Ω resistances in parallel). Now let’s look at a typical filter manufacturer’s datasheet in Figure 5a. In most cases, insertion loss is specified with both the source and load impedances defined as 50Ω/50Ω. Manufacturers often provide both differential- mode (symmetrical) and common-mode (asymmetrical) measurements. Note the term of attenuation is essentially the same as insertion loss with an opposite sign.
22 March 2026
www.electronicsworld.co.uk This naturally raises an important question: Given that these
filters are almost always used to suppress mains noise generated by power supplies, why is a 50Ω/50Ω system used at all? This is a very good question. If we look at older datasheets
(for example, the datasheet from Schaffner – now TE – shown in Figure 5b), we can see that four different attenuation curves were once provided. The additional curves are arguably much more useful. Over time, however, this level of detail has largely disappeared. It is not difficult to understand why manufacturers have
moved toward publishing only 50Ω/50Ω measurements. • First, the measurement is simple and repeatable. A VNA or impedance analyser inherently operates with a 50Ω source and a 50Ω load, allowing the test to be completed quickly.
• Second, even if manufacturers invested additional effort to characterise filters under multiple impedance conditions, very few customers would be able to interpret the results correctly. Several manufacturers have indicated that the majority of customers would not make use of the additional curves – an observation that aligns closely with what I have seen in the field. Whether we like it or not, this is the reality we are working
with today.
Insertion gain Another important concern is that, if we examine Figure 5b, we can see that at lower frequencies the filter exhibits
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48
