Design Article
A practical guide to low power efficiency measurements
Michael Day and Jatan Naik, Texas Instruments
4/25/2007 12:34 PM EDT
When performing measurements on DC/DC converters utilizing PFM, proper care must be taken to ensure that the measurements are accurate. Due to the nature of a converter operating in the PFM mode, its test setup is different from that of a converter operating in the PWM mode. In fact, an improper test setup can result in incorrect efficiency measurement data that varies considerably from data sheet specifications. This article discusses PFM mode and how it helps to maintain high efficiencies at light loads. It also provides guidelines to assist the engineer in acquiring accurate efficiency measurements
Pulse Frequency Modulation
Pulse Frequency Modulation is a switching method commonly used in DC/DC voltage converters to improve efficiency at light loads. This method is also referred to as burst mode and power save mode (PSM). There is one primary advantage that PSM has over traditional PWM schemes: it reduces the power dissipation of the converter at light loads.
A switching converter has two types of power losses: static and dynamic. Static losses are constant, regardless of load current. Alternatively, dynamic losses increase with load current. An example of a static loss is the quiescent current going into an IC. This current is used to power internal circuitry such as bandgap references, operational amplifiers (op amps), internal clocks, etc. In turn, dynamic losses can be classified by two categories: conduction losses and switching losses. Conduction losses are load dependent and include losses caused by voltage drops across a power supply's power MOSFETs and inductor. Higher load currents result in higher conduction losses. A converter also has frequency dependent switching losses that include the MOSFETs' turn-on and turn-off losses, gate drive losses, and body diode losses that occur each switching cycle. As the name implies, these losses are proportional to the switching frequency. Most of these losses are also dependent on load. Figure 1 shows the static and dynamic power losses for low-power ICs. This figure shows that dynamic losses are dominant at higher output currents, while static losses are dominant at lower output currents.

Figure 1. Comparison of a switcher's static versus dynamic losses
Power Save Modes
In an effort to reduce power loss at light loads, many converters operate in a "power save" mode. This mode utilizes a PFM-type of operation at light load currents. This type of operation uses several power saving schemes to maintain high efficiency at light loads. In contrast to PWM mode, in which the converter is continuously switching, PFM mode allows the converter to switch in short bursts. The TPS62350 from Texas Instruments optimizes efficiency over its full input voltage operating range by changing the load current where it enters PFM mode. The PFM load current threshold is Vin/25 Ω. While in PFM mode, the converter only switches as necessary to service the load and maintain the output voltage. When the output voltage drops below its set point, the IC begins switching. As the IC switches, the output voltage rises. This may take one or several switching cycles. Once the output rises above the regulation threshold, the converter stops switching. The output voltage then drops as the output capacitor supplies the load current. When the output voltage drops below the threshold, the converter starts up and switches again. Significant power savings are achieved during the time the converter is not switching. Figure 2 shows this switching function.

Figure 2. SW node operating in PFM mode
When not switching, the converter significantly reduces its quiescent current by shutting down all unnecessary internal circuitry. The only active internal circuitry is the bandgap reference and a comparator to monitor the output voltage. With no switching, all switching losses go to zero. Most converters operate in discontinuous conduction mode (DCM) while in PFM mode. DCM keeps the inductor current from going negative, which would otherwise generate unnecessary conduction losses in both the inductor and power switches. The effect of these power saving schemes is a significant increase in light load efficiency versus standard PWM operation. Figure 3 shows efficiency in both PWM and PFM modes. PFM mode has a 55 percent increase in efficiency versus PWM mode at 1mA.

Figure 3. Efficiency comparison between PFM and PWM modes performing accurate efficiency measurements
Efficiency Measurements
The power saving benefits of PFM mode are critical to extending the operating times of battery-based applications. However, in order to properly model system efficiency and run times, power supply efficiency in both PWM and PFM modes must be properly measured. When measuring the efficiency of DC/DC converters, it is important to properly connect the voltage and current meters to achieve accurate measurements.
Figure 4 shows the setup that should be used to perform efficiency measurements in PWM mode. This figure shows the critical placement of the voltage and current meters for each measurement. Most lab power supplies display their voltage output setting, but it is important not to use the voltage displayed on the lab power supply for efficiency calculations. Instead, connect a separate voltmeter directly across the inputs of the device under test (DUT). This ensures that the measured voltage is the true voltage at the input of the DUT, and does not include additional voltage drops across the current meter or the wires from the input lab supply. The current meter must be placed between the lab supply and the input voltage measurement location. Similarly, a separate voltmeter must be connected directly across the output of the DUT to properly measure the output voltage values. The output voltage should be measured at the point of regulation on the power supply, not at the load. Note that both the input and output voltages are measured with Kelvin connections on the connector. This eliminates measurement error due to IR drops across the connector. Connecting the output current meter in series with the load as shown in Figure 4 provides the correct load current measurement.
Figure 4. PWM mode efficiency measurement setup
Input Waveform
The same features that result in high PFM mode efficiency also make it more difficult to accurately measure efficiency. In Figure 5, the triangular waveform represents the input current of a converter operating in PFM mode. The converter only pulls current when it is switching. Most digital multimeters do not correctly measure the average input current of a power supply switching in PFM mode. Instead of measuring average current, they measure RMS current, which is always higher than the average current. The exception to this statement is when the waveform is purely DC. The engineer can only obtain accurate efficiency measurement by measuring the average input current. This is easily accomplished by adding a large capacitor to the input of the DUT as shown in Figure 6. The lab power supply now delivers a DC current to the DUT. The average input current to the DUT has not changed. The added capacitor filters the AC component of the current required by the DUT and allows the lab supply to source only the average DC current.
The DC waveform in Figure 5 shows the input current with an additional capacitor across the input of the DUT as shown in Figure 6. Proper placement of the input current meter allows it to accurately measure the average input current. Although the current waveform through the meter is purely DC, the current provided by the added capacitor is similar to the triangular waveform above, without a DC offset. Thus, the role of the capacitor can be viewed as separating the input current into DC and AC. A good starting point for determining the value of the added input capacitor is to make it 20 times larger than the power supply's input capacitor. Measure the lab supply current with a current meter and an oscilloscope to ensure that it is a DC waveform. If it still has an AC component, add additional capacitance. The added capacitors should have a fairly low ESR (<100 mohm).

Figure 5. Input current waveforms
Figure 6. PFM mode efficiency measurement setup
PFM comparison
Using the test setup shown in Figure 4 to measure PFM efficiency may result in incorrect data that varies by as much as 15 percent from the actual efficiency. This disparity is most evident at low input voltages and light load currents. Figure 7 compares efficiency measurements taken with and without the additional input capacitor. The need for the additional input capacitor is evident by the fact that the measurements without this input capacitor average five percent lower efficiency than the measurements with the capacitor.

Figure 7: PFM mode comparison with and without input capacitor
Conclusion
Light load efficiency is critical to extending battery life in portable applications. PFM mode uses several techniques to improve light load efficiency, but the benefits can be masked by incorrectly measuring efficiency under these light load conditions. Care must be taken when measuring the efficiency of DC/DC voltage converters to achieve accurate measurements. The placement of sensing instruments is critical, regardless of whether the converter is operating in PFM or PWM mode. Additionally, a large capacitor should be added across the input of the converter to ensure that PFM mode efficiency measurements are properly taken.
About the Authors
Michael Day manages the Low Power/SWIFT Applications Group at Texas Instruments. He received his BSEE and MSEE in Pulsed Power from Texas Tech University, Lubbock, Texas.
Jatan Naik is an Analog Applications Engineer for Power Management Products at Texas Instruments. He received his BSEE from the University of Texas at Dallas, Texas.



