Measurements reveal lost power

To determine how much power a switching power supply loses in its drive circuit, you need to measure the instantaneous power across the circuit. Making instantaneous power measurements is relatively easy, provided you use the right equipment and the proper measurement setups.

Figure 1. A power MOSFET delivers power to a transformer, which filters the modulated signal and delivers DC to a load.

Figure 2. A DSO can capture voltage across and current through a device and then calculate power and energy. Courtesy of LeCroy.

When fully on or fully off, the power MOSFET dissipates relatively little power. When the device is on, the saturation voltage across it (yellow trace in Figure 2) is quite low relative to the voltage across the device in the off state. In the off state, the current is essentially zero. In between states, though, a device can dissipate considerable power because the device can't instantaneously switch between states (white trace in Figure 2).

To measure a device's instantaneous power consumption, you need a digital oscilloscope (DSO). One scope channel measures the drain-to-source voltage (V_DS) and another channel measures the drain current (I_D). Once you obtain the voltage and current, you can use a DSO's math functions to calculate how much power the device consumes.

Note that the MOSFET in Figure 1 has no direct connection to ground. In such a case, you can't use a standard voltage probe to measure V_DS because a standard probe will short one of the device's terminals to ground. Instead, you need to use a differential voltage probe, which will isolate the scope from ground and provide a signal that represents the difference between the voltages at the probe's tips.

To measure I_D, you need to use a current probe. Current probes let you measure current without breaking the circuit.

a)

b)

Figure 3. a) Differences in scope probe propagation delays produce waveforms that are offset in time. b) A DSO can delay displaying a waveform to compensate for skewing errors. Courtesy of Tektronix.

Make sure you use differential voltage probes and current probes that have the bandwidth you need to make the measurements. The required bandwidth will depend on the power device's transition times. As a rule of thumb, divide 0.35 by the rise time of a signal to calculate the bandwidth you need. A 50-ns transition time requires a bandwidth of just 7 MHz, but a 1-ns time requires 350 MHz.

You need to sample the waveforms fast enough to capture the signal's transitions. To capture a rising edge, you need at least two samples on the edge, preferably more. To capture a 50-ns rising edge (typical of a high-voltage MOSFET), you need a sample rate greater than 40 Msamples/s. In low-power applications, a low-power MOSFET might have a rise time of 1.2 ns. At that speed, you need to sample the power signal faster than 1.67 Gsamples/s.

Keep in mind that many DSOs change their sample rate as you change timebase settings. Make sure your DSO provides sufficient sampling at the settings you'll use. Some DSOs also change their sample rate with the number of channels you use so they will sample each of two channels at half the rate that they sample one channel. Most DSOs display their sample rate, so verify that yours samples fast enough to capture the transitions of your devices under test.

You also must verify that the DSO performs the power calculations on voltage and current signals that are aligned in time. Scope probes add propagation delay to a signal. The difference in delay, or skew, between your voltage probe and current probe can significantly affect a power measurement. (Ref. 1).

a)

b)

Figure 4. a) A hard-switched MOSFET produces a large power spike when switching on and off. b) Soft switching reduces power spikes by forcing VDS to 0 V before the device is switched. Courtesy of Xantrex.

Most DSOs let you digitally delay a signal so you can synchronize, or deskew, the voltage and current measurements. Once the measurements align, you can use the scope to deliver a trace of instantaneous power. Figure 3 shows the effects that skew can have on your measurements. The lower trace in Figure 3a shows a difference in time between two signals. Figure 3b shows that using the scope's deskew feature aligns the signals.

You can easily calibrate a scope to deskew signals. You need a square-wave function generator and a resistive load, or you can use a scope accessory that produces synchronized voltage and current signals. If you use a function generator, select a resistor and a square-wave voltage so that the voltage and resulting current sufficiently drive the scope's voltage and current probes. Then, measure the voltage and current in the load and adjust the scope until the waveforms align. If possible, perform the deskew calibration using the same vertical sensitivity settings that you'll use in your power measurements, because the skew can change when you change vertical or horizontal settings. Once you've aligned the voltage and current waveforms, you're ready to take measurements.

Instantaneous power measurements let you compare how different drive circuits dissipate power. A technique called zero-voltage switching (ZVS) uses four power devices in a bridge circuit instead of using a single device as shown in Figure 1. The bridge circuit can force V_DS to 0 V before a device carrying power to the load changes state (Ref. 2) with respect to time. Figure 4a shows the power loss across an IRFP460 from International Rectifier (El Segundo, CA) when "hard switched" (no ZVS, typical of the circuit in Figure 1). The vertical scale is 320 W/div, which the scope calculates by multiplying voltage and current. You can clearly see the large power spikes. Using the same vertical scale, Figure 4b shows that the ZVS technique significantly reduces the power spikes.

Instantaneous power is but one measurement that you must make on a switching power device. You also must measure energy, which you get by integrating instantaneous power (see the lower trace in Figure 2). In addition, you must measure the device's on resistance (R_DS), which you can calculate from the voltage and current measurements. These measurements help you verify that the switching supply uses the best power MOSFET for the application.