Reducing SMPS Conducted Emissions with Three-Winding Transformers

R. De Leo, R. Elisei, M. Marconi, and V. Mariani Primiani

A software model of transformers using a third winding as a shield shows that the configuration is an effective and inexpensive alternative to foil shielding.

It is well known that switched-mode power supplies (SMPS) are an important source of electromagnetic interference (EMI). To comply with the European Electromagnetic Compatibility Directive, the electronics engineer should make sure to address the EMI problem during the early design stage in order to minimize the number of prototypes needed to reach the ultimate design. Obviously, the reliability of the resulting product depends strongly on the accuracy of the design models for individual components. Investigations into equivalent circuits for transformers are therefore still germane.

Parasitic elements in transformers are responsible for both common-mode and differential-mode emissions.¹ This article focuses on the modeling of transformers in order to predict these emissions from SMPS. In particular, the evaluation version 8.0 of Personal Computer Simulation Program with Integrated Circuit Emphasis (PSpice) circuit simulation software (Cadence Design Systems Inc.; San Jose) was used to perform simulations that took into account parasitic elements of all circuit components. The transformer equivalent-circuit component values were extracted by impedance measurements under several load conditions.²

Besides a traditional two-winding transformer, a three-winding version was analyzed as well. Multiwinding transformers typically are used to achieve several output voltages.³ In the case presented here, the third winding was used to create a shield to reduce the conducted emissions from an SMPS. The shielding effect was obtained through appropriate connection to the primary winding. This is a quick and economical way to generate a shield; production time is not appreciably increased.

The equivalent circuit was optimized for use in the PSpice circuit simulator. The simulation of a flyback SMPS allowed the benefits obtained by using the shielded transformer, with respect to conducted emissions, to be evaluated. All simulations were validated experimentally and shown to be in satisfactory agreement with real-world results.

Traditional Transformer Construction

Figure 1 is a simplified electrical schematic diagram of a typical flyback SMPS in which the stray capacitances of the main components are indicated.

The six-capacitance model used for the transformer can be simplified in modeling this type of SMPS. In particular, with the switching transistor being connected to the common point of Cw1 and Cw2, the capacitances Cw3 and Cw4 can be omitted, along with their effect on Cw1 and Cw2. Moreover, for a typical n:1 turn ratio, Cw6 is very small. Therefore, the final equivalent circuit for the transformer becomes that depicted in Figure 2. Also indicated in the figure are the primary and secondary magnetization and leakage inductances, as well as the winding resistances. Each component can be related directly to the windings’ geometric parameters, making this representation useful for design optimization.

Characterization of the transformer is carried out by means of impedance measurements taken at the winding terminals under various load conditions. In particular, the primary winding input impedance, when the secondary winding is open-terminated, is used to determine the primary stray capacitance, winding inductance, and winding resistance (Cp, Lp, and Rp) values (see Figure 3). The same measurement taken with a short condition on the secondary winding allows the primary leakage inductance Lk to be recovered (see Figure 4).

Similar plots can be obtained for the secondary input impedance, even though some secondary parameters (for example, the stray capacitance) can be disregarded owing to the lower turn number.

The stray capacitances between the two windings, C1 and C2, are directly recovered through measuring the impedance between the two windings in an open-circuit condition.

Shielded Transformer Construction

A way to reduce conducted emissions from SMPS is to insert a shield between the primary and secondary windings. This is typically achieved by placing a metallic foil over the primary winding after applying proper insulation. From a practical point of view, this technique is difficult to adapt to automated production. An alternative is to use another winding, which produces the desired shield very quickly and inexpensively.

Figure 5 diagrams an example of a third winding used as a shield. This extra winding is connected to the primary at only one terminal. In the equivalent circuit now are included the capacitances between the primary and shielding windings, Cs1 and Cs2.

It should be noted that, owing to the connection of the shield, the capacitance Cs2 becomes in parallel to Cp, producing a consistent increment of its value. This leads to a visible reduction in the resonance frequencies of the impedance plot, where Cp plays a role.

When the shield winding is added, the equivalent circuit becomes that shown in Figure 6. Lfr is the inductance of the shielding winding, Rf its resistance, and Cs1 the capacitance now introduced between the primary and the shielding windings. A minimum on the impedance module can be expected when Lfr and Cs1 resonate.

Moreover, because the simple PSpice version used for the simulation does not accept a transformer with several secondary circuits, the inductance is reported here without the mutual coupling and assumes the value where Nf is the number of turns in the shielding winding, and k is the coupling coefficient.

Transformer Modeling

The above analysis is based on an equivalent circuit extracted from measurement. In principle, the transformer equivalent circuit can be determined during the transformer’s early design stage by means of a full electromagnetic modeling. Several software packages now commercially available for calculating the complete electromagnetic field inside and around the transformer work from the winding geometry, the characteristics of the magnetic core material, and the case characteristic and geometry.

Having complete knowledge of the electromagnetic field allows an equivalent circuit to be extracted in terms of lumped elements. Nevertheless, considering that the cost of this type of simulator may be too high for a small or medium-sized factory, and that switching transformers are typically designed via empirical formulas derived from previously assessed experience, the measurement-based approach discussed here seems to be a good compromise for extraction of the transformer circuit.

Experimental Results

Two transformers were built that were both suitable to be inserted in the flyback SMPS that is described later in this article. Transformer A had an E 20–type magnetic core, a primary winding consisting of 48 turns of 0.390-mm-diam wire, 3 mm of insulation between the primary and secondary windings, and a secondary winding of three turns of 0.56-mm-diam wire. The shielded transformer B, besides having the same core and primary and secondary windings as transformer A, had 0.130 mm of insulation between the primary and shielding windings, a shielding winding consisting of 46 turns of 0.224-mm-diam wire, and an insulation layer between the shielding and secondary windings 0.390 mm thick.

Equivalent-Circuit Parameter Extraction. All impedance measurements were performed with a resistance, inductance, and capacitance (RLC) meter with a measurement range of 75 kHz–30 MHz, following a proper calibration via a short-open procedure.

Figure 7 shows the two primary impedance plots—open- and short-circuit conditions on the secondary winding—for transformer A. For completeness, Figure 8 displays the same quantities but for the secondary circuit, applying a short and an open load condition at the primary.

With reference to the earlier discussion of transformer characterization, the values for the components of the equivalent circuit illustrated in Figure 2 can be determined. They are given in Table I.

In a similar way, Figure 9 shows the impedance plots for transformer B, from which the parameter values reported in Table II are recovered.

As expected, transformer B’s first resonance in the open-circuit plot exhibits a lower frequency than that of transformer A, and the same decrement is evident in the first resonance of the short-circuit plot. Both effects are due to the increment of the total Cp. At around 10 MHz, a minimum occurs as a result of the series resonance between Lfr and Cs1.

Equivalent-Circuit Validation. The ability of the recovered equivalent circuits to describe transformer behavior under operating conditions was tested by performing a calculation of the primary input impedance loading the secondary circuit with an arbitrary load. This was very different from the previously used short and open circuits. Specifically, an 11-ž resistor was considered; this was the typical load condition for the transformer when operated in the SMPS discussed in the next section of this article. Obviously, the 11-ž resistor itself exhibits stray elements; therefore, it was accurately measured, recovering the equivalent circuit shown in Figure 10.

PSpice software was used for the simulation, which involved inserting the equivalent circuit of the two transformers loaded by the component of Figure 10, and then calculating the primary input impedance. Figure 11 displays the simulation results for transformer A in comparison with measurements from the actual components. Figure 12 presents the same comparison with respect to transformer B.

It can be seen that the adopted circuit description, and the recovered component values, provide a means to predict quite well the measured transformer behavior.

SMPS Simulation

A final test was carried out. This involved performing a PSpice simulation of the flyback SMPS whose electric scheme is displayed in Figure 13.

The main characteristics of this power supply were an input voltage of 80–240 V ac (50–60 Hz), an output voltage of 5 V dc, an output current of 600 mA, power of 3 VA, and a switching frequency of 100 kHz. Because the purpose was to simulate the conducted emissions of the SMPS, the parasitic elements of each component also were inserted in the circuit following an appropriate measurement of capacitor and inductor impedances with the RLC meter. Moreover, to compare the simulation results with actual measurements carried out on a standard test bench, the equivalent circuit of the line impedance stabilization network (LISN) was inserted in the circuit, as well. The voltage produced across the LISN impedances was computed. Simulations and measurements were performed inserting both transformer A and transformer B. As an example, Figure 14 shows the complete electric scheme used by the PSpice simulator in the case of transformer B.

It is important to note the essential role played by common-mode emissions. Unfortunately, it is difficult—indeed, often impossible—to predict them before the equipment under design is actually constructed, because they are generated by stray capacitances between circuit elements, which are exposed to fast voltage changes, and the reference ground, the common point of the LISN.⁴ Therefore, after the printed circuit board of the SMPS was constructed and installed in the conducted-emissions test bench, these parasitic capacitances toward ground were measured and inserted in the simulated circuit (see Cfb1 and Cfb2 in Figure 14).

These capacitances can also be roughly estimated by means of standard formulas if the geometry—PCB traces, conductor diameter and length, distances from the reference ground plane—is known, even if the SMPS is in an early prototype form.⁵ In the present case, the estimation yields values ranging from 10 to 50 pF; however, owing to the importance of these capacitances for common-mode noise propagation, it is better to measure them accurately after generation of the PCB and its mounting on the test bench. The measured values in the test under discussion were 9.4 pF for the primary circuit (Cfb1) and 113 pF for the secondary circuit (Cfb2), which accords with the fact that the secondary-circuit traces are wider than the primary ones due to the higher current.

Figure 15 reports the simulated and measured conducted emissions produced by the SMPS with transformer A. The same comparison with respect to transformer B appears in Figure 16. In addition, both figures show the limit fixed by the standard EN 61000-6-3.⁶

In the instructional version of PSpice software that was used, simulations are restricted to a maximum 10 MHz and the EMI filter is not inserted in the circuit in order to limit the number of components. Nevertheless, simulation results show a satisfactory agreement with measurements. The results highlight the effectiveness of using the third winding as a shield, moreover, particularly beyond 5 MHz.

It is interesting to compare the conducted-emissions results obtained with those achieved when a metallic foil was used for the shield. Figure 17 presents the experimental comparison: the emissions are quite similar up to 10 MHz, but above that level, the winding method is better for the SMPS under consideration.

Conclusion

The use of a shield between the primary and secondary windings of a transformer reduces conducted emissions in the higher frequency range, specifically, above 5 MHz. Employing a third winding as the shield makes transformer construction easier and allows for simpler design modeling than metallic foils used as shielding. Moreover, experimental investigations have highlighted the fact that this solution reduces emissions more effectively than the classic foil approach.

The equivalent circuit chosen for the transformers, together with the parameter extraction technique described here, is suitable for predicting the measured conducted emissions of a typical SMPS. For this, it is sufficient to insert the LISN equivalent circuit in the scheme and to add adequate common-mode capacitances, whose values can be estimated from measurements.

References

1. JC Fluke, Controlling Conducted Emission by Design (New York: Van Nostrand Reinhold, 1991), ch. 10.
2. B Cogitore, JP Kéradec, and J Barbaroux, “The Two-Winding Transformer: An Experimental Method to Obtain a Wide Frequency Range Equivalent Circuit,” IEEE Transactions on Instrumentation and Measurement 43, no. 2 (1994): 364–371.
3. A Schellmanns, K Berrouche, and JP Kéradec, “Multiwinding Transformers: A Successive Refinement Method to Characterize a General Equivalent Circuit,” in Proceedings of the IEEE Instrumentation and Measurement Technology Conference (Piscataway, NJ: Institute of Electrical and Electronics Engineers, 1998), 717–722.
4. MJ Nave, Power Line Filter Design for Switched-Mode Power Supplies (New York: Van Nostrand Reinhold, 1991), ch. 2.6 and ch. 5.
5. CS Walker, Capacitance, Inductance, and Crosstalk Analysis (Norwood, MA: Artech House, 1990), ch. 2.2.
6. EN 61000-6-3:2001, “Electromagnetic compatibility (EMC)—Part 6-3: Generic standards—Emission standard for residential, commercial and light-industrial environments” (Brussels: CENELEC, 2001).

R. De Leo and V. Mariani Primiani work in the Dipartimento di Elettromagnetismo e Bioingegneria at the Università Politecnica delle Marche (Ancona, Italy). De Leo can be contacted via e-mail at roberto.deleo@unian.it. Primiani can be contacted at valter.mariani@univpm.it. R. Elisei and M. Marconi both work for Comelit S.r.l. (Ancona, Italy).