Thursday, June 21, 2018
Home/Current Issue >  Industry Articles >  Test and Measurement > 

Measuring Losses in Switch-Mode Power Supplies
Power test setup with HDO oscilloscope.

Switch-mode power supplies are known for efficiencies that can run into the high 90 percent range, a far cry from the 50-60 percent efficiencies one may expect from the older technology of linear power supplies. Power-supply efficiency is a measure of how much energy is wasted between the unit's input and output. There are other gains to be expected from switch-mode supplies as well, such as higher power densities. If a given application can tolerate the increased noise that switch-mode supplies typically generate and/or does not require a fast transient response, then the switch-mode supply is often the better bet.

Switch-mode power supply efficiency is important for two key reasons. For one, a more efficient unit means more potential size and weight savings, which can be critical in the design of portable systems. For another, turn-on, turn-off, and conduction losses within the supply are converted into heat, which can compromise the health and longevity of the power supply itself as well as that of the system it powers. Thus, regardless of a power supply's efficiency, it's important to know how much energy is being lost in the transition as the output transistor turns on and off, as well as how much is lost in conduction.

In virtually all bench scenarios, measurements of power-supply losses are made with an oscilloscope. At the outset, there are some important considerations with regard to the test equipment itself. Most general-purpose oscilloscopes are useful for ground-referenced measurements, which mean that the oscilloscope probe's ground lead is connected to the instrument's case. The case is connected to earth ground through the ground lead in the oscilloscope's power cable.

The resulting single-ended probe is prone to ground-loop effects that can cause the voltages at the oscilloscope's BNC input connector to be unequal to the voltages at the probe tip. This becomes a problem particularly when measuring low-amplitude signals such as power-supply losses, because the noise and voltage gradients in the ground-distribution system can be as large, or even larger, than the signal being measured.

No Ground Reference
Then there's the problem of measuring signals that are not referenced to ground, or "floating." One solution that users have employed is to simply cut the ground lead in the power cord, allowing the case to "float" to the voltage present on the probe's ground lead and breaking any ground loop with the circuit under test. This is a potentially dangerous practice, exposing the user to shock hazards, not to mention incorrect measurements.

A better approach is a probing system incorporating a true differential amplifier, which will give users the best measurement quality for signals which are not referenced to ground. A differential amplifier's output is grounded, which means that the oscilloscope also is grounded, making for a safer situation for users. Most importantly, differential amplifiers have a large common-mode range, permitting measurement of small differential signals relative to large common-mode voltages. In a true differential system, the two input paths to the amplifier are precisely matched, which further improves common-mode rejection and noise reduction. An example of such a true differential amplifier is Teledyne LeCroy's DA1855A, which offers a common-mode rejection ratio of 100,000:1.

Eliminating Error Sources
Another equipment-related consideration when measuring power-supply losses is to carefully eliminate sources of potential error. This includes, among other things, eliminating any timing skew between voltage and current probes. A helpful accessory is a deskew calibration source. A second key consideration is elimination of DC offset errors by auto zeroing the probes.

With potential equipment-related foibles addressed, let's take a closer look at what we mean by the losses incurred in a switch-mode power supply. In such circuits, a switching transistor (usually a field-effect transistor or FET) is switched on and off at a fixed frequency. Each time the transistor switches on to conduct current, a given amount of energy is transferred to the output. The output voltage is measured internally and that value is passed to a controller that regulates the supply's duty cycle. If the voltage falls below the desired value, the duty cycle is increased, and if it rises too high, it is decreased, all in the interest of maintaining a steady output-voltage value.

When the output transistor is conducting current, there is a small voltage drop across the small channel resistance RDS(ON). Meanwhile, the transistor feeds an inductor in series. The current in that inductor cannot change instantaneously when the transistor switches on, but rather ramps up over some period of time. When the transistor turns off, energy stored in the inductor bleeds off into the secondary, which constitutes the mechanism of energy transfer to downstream circuitry.

During the time period in which the transistor is turned off, no current flows, but high voltages are present across it and very little power is dissipated. While the transistor is conducting, voltage across it is very low, perhaps 1 to 2 V. This is the drop across RDS(ON), known as the IR drop, which, when multiplied by the current, gives us the conduction loss.

Transition Power Losses
Power dissipated in the transistor is obviously low when it is in the off state, and is also low while it is conducting. But when it's transitioning in state from off to on, and then from on to off, the voltage and current both change at a finite rate. Power is dissipated by the device and that's power lost as heat.

With older versions of power analysis software, measurements of the turn-on, turn-off, and conduction losses required a labor-intensive setup that involved gated parameters and cursors. Calculation of the area underneath the voltage and current curves depended on the placement of those parameter gates. Those area calculations are what the oscilloscope used to determine the losses. Today, there are oscilloscopes that can be equipped with power analysis software that automates the measurement process. Teledyne LeCroy's HDO high definition oscilloscope is one such instrument.

A typical screen setup on an HDO oscilloscope for measurement and analysis of switch-mode power supply losses might entail three grids. The top grid shows the voltage waveform, in which it can be seen that voltages are very low during the conducting portion of a cycle and very high when the transistor is in the off state. A second grid displays the current waveform, where users can see how the current ramps up during the period in which the series inductor is charged. This period corresponds to the power spike that appears in the third grid, which tracks device power dissipation. The spike is the result of the current for the inductor being dissipated in the channel's resistance.

Improved Measuring Method
The improvement in the newer method lies in its relabeling and separating out of the various parts of the transistor's switching cycle. Rather than using cursors and gated measurements to arrive at turn-on, turn-off, and conduction losses, they are measured independently and then read out independently. No longer does the user have to struggle with setting up parameter gates. It's a matter of probing the circuit and directly reading out the results.

Contact: Teledyne LeCroy, 700 Chestnut Ridge Road, Chestnut Ridge, NY 10977 800-553-2769 or 845-425-2000 fax: 845-578-5985 E-mail: Web:

search login