In every product, there is a balance to be made between cost and performance. Cell phone adapters are a prime example. One trade-off is in the design of the electromagnetic interference (EMI) filter where component cost must be minimized while the product has to pass strict EMI emission requirements. Figure 1 shows an example EMI filter and power stage for an adapter.
Figure 1: Common-mode inductor (L1) may not be needed at low power. (View full-size image.)
As discussed in Power Tip 47 and 48, common-mode currents are generated by the high-voltage switching waveforms applied across the stray capacitances. These capacitances can be quite easy to visualize, such as the primary-to-secondary transformer capacitance, C_Stray2.
Or they may not so easy to visualize, such as the stray capacitance from the transformer core to chassis represented by C_Stray1. These common currents flow through the chassis connection whether it is an intentional physical connection to chassis or just capacitive coupling. As the currents complete their path through the input source, they can cause a product to fail EMI testing.
The typical approach to reducing common-mode emissions is to return common-mode currents in the transformer of Figure 1 through C1 and to add a common-mode inductor, L1, to limit current flow. The challenge is that the common-mode inductor adds cost and size to the product, which is particularly undesirable to low-power, high-volume products like cell phone chargers.
The following describes a series of incremental changes to the EMI filtering of a real design with the goal of eliminating the common-mode inductor. Figure 2 is the baseline EMI measurement which displays the CISPR class B limits and the first two measurements.
Figure 2: C1=4700 pF makes dramatic improvement in EMI. (View full-size image.)
We removed the common-mode inductor (L1) and common-mode capacitor (C1) and made the measurements. We measured emissions of over 30 dBuV out of spec due to common-mode current through the transformer capacitance C_Stray1. This current continued into the secondary circuits and through stray capacitance into the chassis.
With C1 equal to 4700 pF, we measured a significant reduction in emissions of 30 dBuV, as shown on the plot. This improvement is due to the return of the common-mode currents through the added capacitance (C1). Adding C1 also changed the frequency at which the emissions peaked. These emissions are generated by the transformer magnetizing inductance resonating with the total stray capacitance on the drain of the MOSFET.
With C1 open and no secondary chassis connection, there is significant impedance in series with CStray,2 so it did not add much to the total stray capacitance on the drain. With C1 in the circuit, CStray2 added to the total stray capacitance and reduced the resonant frequency.
Even with C1 in the circuit, we did not pass emissions. This is largely due to C_Stray1, which includes capacitive coupling to chassis of everything connected or capacitively coupled to the MOSFET drain. This includes traces, snubber components, transformer windings and even the transformer core, which is capacitively coupled to the drain.
Next we reduced the noise source within our circuit (Figure 3), which displays two time domain voltage measurements of Q1 drain. The larger ringing trace is the baseline and the smaller is with a damping circuit connected to the secondary of the transformer (Figure 1).
Figure 3: Damping T1 reduces drain voltage ringing that generates EMI. (View full-size image.)
In the time domain, it is clear that emissions are reduced. Figure 4 shows the measured emissions, which have been reduced by about 6 dB and the peak has been shifted to a lower frequency. Further damping further reduces emissions; however, at this point we begin to impact the efficiency of the power supply.
Figure 4: Reduced ringing of Q1 drain further reduces EMI emissions. (View full-size image.)
Shielding of the transformer can be beneficial as well. The idea is to put a shield between the high-voltage winding and the secondary circuits and to connect the shield to the return or high side of the primary circuits. This prevents common-mode currents from flowing into the secondary circuit and then into the chassis. After implementing a shield, we meet the emissions requirements achieving another 6 dB improvement.
To summarize, common-mode emissions are generated by a high-voltage waveform coupling through a capacitance, which eventually makes it to chassis and returns through the system's input power connection. The obvious ways to reduce emissions are to reduce the amplitude of the switching voltage waveform, to divert the current with a shield or capacitor, and/or to add a series impedance. We applied the first two techniques in the paper to demonstrate we can build a low-power, AC input adapter to meet EMI emissions without a common-mode inductor.
I would like to thank GCi Technologies in support of this project with their numerous transformer iterations that allowed us to pass emission requirements. Please join us for the next Power Tip where we will look at synchronous rectifiers in discontinuous flybacks. See the index of all Power Tips articles (Word .doc file).
Check out TI PowerLab Notes for a designer's perspective on his power supply designs. For more information about this and other power solutions, visit: www.ti.com/power-ca.