Today most of engineers use oscilloscope as the preferred measurement tool of choice when it comes to debugging and analyzing switching power supplies. In this session we will learn about some basics of power measurements, key measurements typically performed in designing a switching power supply and how Agilent s new power measurements and analysis software can help characterize the switching power supplies automatically, consistently and fast. 1
Here is the agenda for the day. 2
What do these electronic devices have in common? Power supply The power supply is integral to every type of electronic products today, and the switching mode power supply has become the dominant architecture in computers, networking devices, test instruments and communication devices. 3
This slide talks about the trends in switching power supplies market place that drive enhanced capabilities in test solutions. 1. Cost of switching power supplies continues to drop. Average cost of modern AC/DC power supplies is coming down to 10 cents/w these days. 2. A modern switching power supply s s efficiency (Pout/Pin) ranges from 65% to 95%. Power efficiency and power density of a SMPS is ever getting higher to generate more power per cubic space. 3. Modern electronic devices require higher reliability to ensure that the devices operate flawlessly. 4. In 2001, the European Union put into effect the standard IEC/EN61000-3-2 to set limits on the harmonics of the AC input current up to the 40 th harmonic for equipment above 75 watts. Designers now have a need to perform pre-compliance testing to the CE standard. 4
5
Power is defined as the rate of flow of energy past a given point. In alternating current circuits, voltage and current only remain in phase if the load is purely resistive. When this happens the power is said to be 'real power'. If instead the load is purely reactive (either Capacitive or Inductive), all of the power is reflected back to the generator as the phase cycles. The load is said to draw zero real power, instead it draws only 'reactive power'. If a load is both resistive and reactive, its will have both real and reactive power, resulting in total amount of power called the 'apparent power'. 6
The job of all power supplies is to produce well-regulated and low noise DC output power from an AC input rail. There are two main types of power supplies: Linear power supplies, also known as series-pass power supplies, operate within the transistor s linear region. These power supplies offer lower noise and EMI compared to their SMPS cousins. Due to lower noise, less filtering is required at the output. However, these power supplies have their tradeoffs. They tend to be less efficient than SMPS resulting in a greater degree of power loss. Switching Mode Power Supplies (SMPS) also use transistors in their designs typically MOSFETs. However, instead of operating the transistor solely in its linear region, an SMPS quickly switches the transistor from its ON, or conducting state, to its OFF, or non-conducting state. As you can imagine, with a high h degree of switching comes a higher h degree of noise and EMI. However, the benefit of this architecture is increased efficiency which means that there is lower power loss compared to linear power supplies. 7
This is a basic block diagram of a switching mode power supply. The AC input first passes through a rectifier, which converts the signal from alternating current (AC) to direct current (DC). A full-wave rectifier converts the whole of the input waveform to one of constant polarity (positive or negative) at its output by reversing the negative (or positive) portions of the alternating current waveform. The positive (or negative) portions thus combine with the reversed negative (or positive) portions to produce an entirely positive (or negative) voltage/current waveform. However, this still does not result in a constant-level DC voltage signal. So, in order to get to a constant-level DC voltage, a smoothing circuit or filtering stage is required. After the signal has been filtered, it passes to the switching device or power transistor stage. During this stage, the power transistor rapidly switches between its ON and OFF states. While in the ON state, the transistor is conducting (current is flowing). While in the OFF state, the transistor is non-conducting (there is no current flow). This effectively chops up the signal into discrete chunks. The transistor stage steps-up the signal resulting in a higher voltage level than at the input or steps down the signal (resulting in a lower voltage level than at the input). The signal is then further filtered to remove ripple or EMI that was generated in the transistor stage. The goal is to have a wellregulated DC supply at the output of the power supply. 8
So, how does switching lead to greater power efficiency? Take, for example, a linear power supply. By operating the transistor in its linear region, the transistor basically operates as a variable resistance. The instantaneous power dissipated by any device is equal to the product of V times I, where V is the voltage drop across the transistor, and I is the current flow through it. In the linear region of operation, the transistor is in a partially-conducting state. The voltage drop across the transistor is Vin Vout. Thus, there is always some power dissipation i associated with linear power supplies. In a SMPS, the transistor is switched between two states the ON state, or conducting state, and the OFF state, or non-conducting state. During either of these two states, the instantaneous power dissipated is ideally always equal to zero (in the ON state, V = 0, resulting in P(loss)= 0; in the OFF state, I = 0, resulting in P = 0). Of course, no transistors are perfect, so there is always some loss but the loss is considerably less than what you would find in a linear power supply. Note that increasing the speed of the switching results in lower P(loss), but may result in greater EMI transferred to the output. MOSFETs, the most popular kind of transistors used in SMPS applications, have certain types of losses associated with them switching losses, conduction losses, and gate charge losses. We ll talk more about these on the next slide. 9
10
A typical Block diagram for the AC/DC switch mode power supply is shown here. The blocks in yellow represents the type of analysis designer do while designing and testing a switch mode power supply. 11
The commonly used power devices in a switch mode power supply are MOSFET, isolation transformer, and inductors. These devices dissipate power during switching and conduction time, causing less than ideal efficiency. To improve the power efficiency of the power supply, designers characterize for instantaneous power loss at the switching device (called switching loss). They also measure Dynamic ON resistance, to see power loss while MOSFET is conducting (called conduction loss). Gate charge loss also is the source of power losses which is not measured in Agilent s power measurement application. 12
To see the reliability of the power supply it is very important for power supply designers to measure the power loss at MOSFET during dynamic load changes. On the picture shown here, the channel 1 (yellow) s the voltage across drain and source of the power MOSFET, channel 2 (Green) is the source current, and the purple waveform in the bottom is the product of voltage and current, indicating the power dissipation at the device. 13
The voltage across the switching device will be high while the device is off, and will be low (V saturation) during the conduction time (On state). During the Off state of the device, there is no current. However, at conduction time the current reaches its maximum. If we look at the power waveform, the maximum instantaneous power loss occurs during transitions (called switching loss). Power loss during the conduction time is called conduction loss. The power loss in the entire cycle is called cycle power loss. 14
The dynamic ON resistance gives us an indication as to the total conduction losses in a power device. The ON resistance, which is the resistance between the drain and the source of the MOSFET, is measured during the ON state of the transistor. Remember, a transistor in an ON state will ideally be conducting, and have zero voltage drop across the drain and source. However, given that we live in a non-ideal world, there is always some small voltage drop. The degree of Rds gives us an idea as to the degree of power loss during the conducting state of a transistor. The dynamic ON resistance measurement works by looking for the point in the waveform where the voltage is close to zero. The software applies a moving average smoothing filter (256 pts:1pt) to reduce the impact of white noise. This enables the software to effectively average out the noise introduced by the scope and the probe enabling a more precise measurement. 15
Since we are on the topic of ON resistance, let s discuss the best way to make the measurement. Traditionally, an engineer typically zooms in on the Vds signal with the goal of obtaining very fine resolution on the voltage level, Vds. The drawback of this method is that you run the risk of placing the oscilloscope into an overdrive state, where the front-end amplifiers are saturated. When this happens, the oscilloscope will display signal distortion which results in inaccurate measurements for a certain period of time before it recovers from the overdrive condition. Some oscilloscopes are known to have better overdrive recovery than other oscilloscopes. However, all oscilloscopes are vulnerable to an overdrive state, so it is best to avoid this condition if possible. In this example where Agilent oscilloscope has +/-8 div of vertical dynamic range, you can see on the right that the waveform exceeding the input dynamic range begins to distort. 16
To avoid placing the oscilloscope into overdrive, Agilent s Power Measurement Application acquires the signal on-screen, optimizing for resolution by using the full dynamic range, and then transfers the data to the application for further processing. In the application, a smoothing average filter (256pts:1pt) is applied to both the Vds and Id signals to improve resolution. The dynamic ON resistance is then calculated between the 25% and 75% points of each pulse period. Rds = Vds/Id 17
Every switching device has a maximum voltage and current rating specified by the manufacturer, displayed on its technical application note or data sheet. The safe operating area test determines how much current can run through the transistor at a given voltage level. The SOA limit is a 5-point polygon and varies from transistor to transistor. Switching device data sheets should provide a safe operating area plot for each device. Points that fall outside of the polygon violate the Safe Operating Area and are indicated d in red. The equivalent violation points in the Vds vs. time and Id vs. time plots are also displayed in red. Here we are viewing 100 switching cycles during >1.2 msec of time and ensure that the transistor is operating well within the maximum voltage and current ratings. 18
Another key measurement that SMPS designers are concerned about is measuring noise and ripple. As we talked about earlier, switching power supplies generate more noise and EMI than their linear power supply cousins. This noise and ripple can appear at the output of the power supply and depending on the characteristics of the end product could cause damage. Noise and ripple is typically very small relative to the output signal only in the mv range. 19
Again, we are going to talk about how Agilent makes the measurement with the Power Measurement Application. First, it is important that the customer use the right probe for the application. Because we are measuring only mv that are riding on a DC output, a high voltage differential probe will not provide the resolution or accuracy we need for the measurement. Therefore, it is important that a 1:1 passive probe, or even a 10:1 passive probe, is used for the measurement. This will provide the best resolution on the small AC ripple signal. Secondarily, in order to remove the DC offset from the signal, the oscilloscope is placed into AC coupled mode. This allows us to maximize the vertical resolution on the ripple Lastly, some filtering may be required. Many power supply designers use scope s built-in 20 MHz or 25 MHz low pass filter to cut off unwanted high frequency noise coupled to the output. 20
Transient analysis test is performed using a step function change in the output load current and monitoring the time required to settle within a specific time. This test is often performed from 50% load to full load in both directions of nominal input line conditions. 21
Often designers have a need to measure turn-on time and turn-off time of switching power supply. Turn-on time is the time taken to get the output voltage of the power supply after the line input voltage is applied. Turn-off time is the time taken to get the output voltage of the power supply to turn off after the line input voltage is applied. 22
The Modulation Analysis measurement subgroup focuses on a set of measurements specific to the feedback loop (also called control loop ) of the power supply. It is the feedback loop that controls the switching rate of the power supply in order to generate a well-regulated DC output. Modulation analysis can also be used to verify power supply stability under changing load or line conditions. 23
In the switching power supply, chopping the input voltage controls output power. Chopping the input voltage is achieved by PWM signal. Designers often need to see the on-time and off-time information of the PWM signal which is very difficult to visualize because the information bandwidth is much slower than the pulse width switching frequency. This specific measurement trends the frequency, period, duty cycle and pulse width variation for the input signal over time. Another benefit of Agilent power measurement software is that t you can slit the measurement window in to one, two, four graphs or four quadrant charts and display each different measurements on each window. 24
At power-on, the input current absorbed by the power supply has a spike which should not exceed the maximum allowable input current. The green trace in the screen capture shows the example of inrush current of an AC/DC power supply, shown together with the line voltage waveform. The power measurement software allows you to automatically measure the instantaneous value of the input surge current to a power supply when AC power is first applied. 25
In 2001, the European Union put into effect the standard IEC/EN61000-3-2 to set limits on the harmonics of the AC input current up to the 40 th harmonic for equipment above 75 watts. The standard defines four classes of equipment depending on its type and current waveform. The most rigorous limits are set for personal computers, computer monitors, and TV receivers. Individual harmonic content is compared against the IEC 61000 3-2 standard for general power supplies and the RTCA DO-160E standard for airborne equipment. For the FFT plot, the Blackman Harris window is used due to its minimal spectral leakage characteristic. Once the FFT is performed, the software keeps the first 45 harmonics only, because most standards only require the analysis of the first 40 harmonics. 26
This measurement analyzes the power quality of the AC input current. Some AC current may flow back into and out of the load without delivering energy. This current is called reactive or harmonic current, and gives rise to an apparent power which is larger than the actual power consumed. Power quality is gauged by the following: Apparent power: The portion of the power flow due to stored energy, which returns to the source in each cycle Real power: The portion of the power flow that, averaged over a complete cycle of the AC waveform, results in a net transfer of energy in one direction Power factor: Ratio of the actual power to apparent power Crest factor: Ratio between the instantaneous peak current and voltage required by the load and the RMS current and voltage. 27
Agilent s U1881A/U1882A power measurement software makes it easy to analyze the line power. It simplifies the calculations of real power, apparent power, and power factor (real power/apparent power) by eliminating the need to set up math traces and parameter math or using an expensive AC power analyzer or AC power source. 28
29
Agilent s power measurement application provides automatic, consistent and fast characterization of SMPS at a price comparable to or lower than our competitors offerings. 30
Quick over of Agilent s power measurement applications There are two use models 6000/7000 InfiniiVision scope with the software running on an external PC. The scope and PC can be connected over USB, LAN or GPIB (6000 only). Or the software can be run inside the windows based Infiniium 8000 scope. Note that we currently do not support high-end 80000 or 90000 series. The power measurement software also support off-line analysis mode as well as on-line mode. With off-line mode you can make full power measurements off-line with previously stored on-line measurement data. 31
Here is the typical configuration of power measurement systems consisting of a scope (6000), power measurement software running on an external PC, high voltage differential probe, current probe, and SMPS DUT. 32
The power measurement software user interface is comprised of a number of blocks. -Test selection menu to choose the measurement - vertical task tabs to set up probe configurations, measurement time scales, connectivity mode (on-line, off-line) etc - measurement listers to display various measurement results - waveform graticules to show scope s s waveform display and allow user to pan and zoom on a particular area of interest, or gate the area of interest and run the measurement on that particular gated area. The Waveform area can be splitted in to a single, dual, quadruple or four quadrant windows. 33
Voltage and current probe deskew is critical for making accurate power loss measurements. Any skew between the voltage and current probes can add or subtract to the real power loss, giving you a faulty, inaccurate measurement. It is recommended that deskew is performed for each set of voltage and current probes being used, and prior to running the Power Device Analysis and Input Line Analysis tests. The software will automatically prompt the user to deskew their probes the first ttime the Power Device Analysis or Input tline Analysis tests t are activated. The software will assume the same deskew values for each test that is run in a single open instance of the application. If the application is shut down and re-opened, the application will prompt the user for an updated deskew measurement. 34
Here is the list of other probes and accessories needed to perform power measurement applications. Agilent provides the variety of current probes and voltage probes needed to perform power measurements. 35
Agilent s power measurement applications provide the most measurement features than any other industry s scope based power measurement applications. 36
Why buy Agilent s power measurement solutions? Lower entry price. The software works with a broad range of 100MHz 1GHz DSOs and MSOs. Our power measurement solution hits much lower entry prices vs competitive scopes with comparable feature sets. Automatic test configuration with your choice of multiple power measurements in a single display. More measurements. More features such as inrush current, load transient response measurement utilize the benefit of MegaZoom deep memory. 37
38
39