EMI from offline switching power supplies typically causes all sorts of problems for power-supply designers. You may need a large EMI filter to meet FCC emission requirements. Switchers for high efficiency produce high-frequency switching noise that can propagate through the rest of the system and cause problems. Board layout is critical, requiring considerable experimentation, even for experienced designers. The low-noise circuit in Figure 1 significantly reduces the complexity of these issues by continuous, closed-loop control of the voltage and current slew rates. High-frequency noise suppression is particularly important for medical devices because they don't require the ac-line-to-earth ground capacitors ("Y" capacitors) that typically suppress this noise. The absence of these capacitors allows medical devices to easily comply with the more stringent low-leakage-current health-care specifications of UL544, UL2601, and CSA22.2.

Figure 1 shows a 30W (12V output at 2.5A) offline power supply. IC₁, an LT1738 low-noise switching regulator in a flyback topology, drives Q₁ and continuously controls the current slew using the resistor at the R_CSL pin. The IC controls the voltage slew using the resistor at the R_VSL pin and the capacitance at the CAP pin. IC₂, an LT1431 programmable reference, and the optocoupler close the isolated loop back to the LT1738. The circuit achieves current limit by sensing the current through a 68-mΩ resistor at the CS pin. Q₂, Q₃, and their associated circuitry provide undervoltage lockout with hysteresis. During start-up, the pin stays low until C₅ charges to 12V via R₁. The LT1738 then turns on and subsequently obtains most of its operating power from T₁'s auxiliary winding. The feedback goes directly to the LT1738's V_C pin rather than to the F_B pin because the optocoupler provides the feedback gain that the LT1738's internal feedback amplifier typically provides. C₆ and L₁ attenuate the low-frequency harmonics of the LT1738 switching frequency.

You can see the benefits of the circuit by measuring its ac-line-conducted EMI and then comparing these measured results with those for basically the same circuit with the LT1738 replaced with a generic switcher. The only circuit-parameter difference is that, unlike the LT1738, the generic switcher doesn't actively control the switching current and voltage slew rates. Figure 2 shows the frequency spectra for both circuits. You can see by the respective frequency spectra that the LT1738-based circuit generates emissions well within FCC Class B requirements, whereas the circuit with the generic part results in emissions that exceeds FCC Class B allowable emissions by a significant margin.

Another benefit of the circuit in Figure 1 is that the output voltage noise comprises the fundamental ripple with practically no high-frequency components. You can attenuate this ripple voltage if desired to less than 300 µV using a 100-µH, 100-µF LC filter on the output. The generic switcher, on the other hand, produces more output noise because the high-frequency noise passes to the output with little attenuation through the parasitic capacitance of the output filter's inductor. The circuit in Figure 1 minimizes noise and EMI by controlling the voltage and current slew rates of the external n-channel MOSFET. This circuit is well-suited for offline power supplies in medical devices because the absence of Y capacitors results in low leakage current to earth ground in compliance with health-care specifications.

Is this the best Design Idea in this issue? Select at www.edn.com.