Powering the Generation of New Products

“Spoiled for choice” is an understatement when it comes to DC power supplies and AC sources. There literally are hundreds of models and options to choose from. Output voltage and current generally top the list of requirements, but a great many factors influence the complete specification.

the two comparison charts accompanying this article.

Some DC supplies, such as Keithley’s 4500-MTS Multichannel Test System provide many independent voltage and current sources that can be programmed to test several DUTs in parallel. This type of product combines power-supply capability with measurement and analysis to perform detailed pulse or level testing.

In another case, such as the development of a gigahertz microprocessor PCB, this capability would be meaningless. Fast microprocessors have very high current levels and di/dt transitions associated with their operation. The most suitable power supplies tend to use multiphase rectification to ensure low output ripple while maintaining fast transient recovery.

Regardless of the application, you may be governed by EN 61000-3-2 current harmonic limits, which usually are met through some form of power factor correction (PFC). The intention of this and similar IEC standards is to reduce AC powersource pollution caused by AC current harmonics. Especially in capacitor-input supplies with small diode conduction angles, harmonics can be high and the power factor (PF) low.

Although reducing AC waveform distortion is important, PFC also has the very practical benefit of maximizing the amount of useful power that can be drawn from a limited-capacity supply. For example, 115-V single-phase AC power often is limited to 15 A per circuit. With PFC, a power supply can use nearly the entire 1,725 VA. If voltage and current are in phase giving a PF of 1.0, the power supply looks like a toaster to the AC input, and 1,725 W can be consumed. Without PFC, the PF of a conventional capacitor-input power supply could be 0.6 or lower. In this case, only 1,035 W of real power are available. The remaining 690 VA appear as reactive power—important to the utility company supplying it but otherwise wasted capacity that a low-PF supply can’t use.

PFC Explained

Readers with an electrical or electronics background will remember that PF = cos φ, where φ represents the phase angle between voltage and current. In simple applications involving inductive loads, such as in a factory with several large motors, the voltage and current waveforms remain sinusoidal but out of phase. This situation can be corrected by adding capacitance to counteract the effect of the inductive motors, and in fact, many companies supply large capacitor banks for this purpose.

Modern power supplies used for electronic applications present a more complex situation because the current waveform is no longer sinusoidal. There is no simple phase relationship to correct, and the more general definition of PF must be used: PF = W/VA regardless of the wave shape. For low-wattage supplies, passive input filters can be used to extend the diode conduction angle and reduce the level of harmonics. For higher power supplies, passive input filters become large and expensive, therefore, active PFC is more practical.

Active PFC Topology

Most power supply manufacturers use a separate high-frequency boost-type switching circuit to force the average input current to be in phase with the input voltage.

As shown in Figure 1a, the boost configuration of an inductor, switch, and capacitor has inherent EMI filtering properties because the inductor is placed ahead of the switch.1 On the other hand, this topology requires the DC output to be higher than the peak input. A common output voltage is 380 to 400 V to allow operation from 115- or 230-V inputs. Considering Figure 1b, the boost-converter topology allows the input current and the inductor current to remain continuous throughout the entire line cycle.1 This is not the case with buck or buck-boost topologies that place the switch in series with the input: although the inductor current may be continuous, the input current is not. With these configurations, an EMI filter must cope with large amounts of high-frequency input-current noise.

PFC Control

Figure 2 shows the general control scheme used in many boost-type PFC circuits.1 The outer voltage loop has a bandwidth about 10% of 2× the line frequency or 10 to 12 Hz. The output voltage is compared to a DC reference, the error voltage being one input to a multiplier. The other input is a fraction of the rectified input sine wave. A scaled version of the rectified sine wave appears at the multiplier output and acts as the reference for the faster current control loop. If the average current can be forced to follow the shape of the input voltage, a PF near unity will result.

The high-bandwidth current loop drives a pulse-width modulation (PWM) stage to control the current value.

Although this description is correct, there are several details that vary among manufacturers and affect characteristics such as voltage stress across the switching element, efficiency, operating frequency, and duty cycle.

For example, operating in the critical conduction mode (CCM) turns on the switch when the inductor current reaches zero. CCM describes the boundary condition between truly continuous inductor current and the discontinuous mode of operation.

To View DC Supplies Chart Click Here

To View Features Chart Click Here

The switch is turned on for a fixed time, causing the inductor current to ramp up to a peak current proportional to the rectified sinusoidal voltage. After the switch is turned off, the inductor discharges at a rate determined by the rms AC input voltage, the load current, and the instantaneous voltage within the line cycle. Because of these dependencies, the switching frequency for a CCM PFC circuit varies widely, making EMI filtering more difficult.

Against this drawback, CCM operation has the benefits of soft and nearly lossless turn-on switching, ease of stabilization, and control simplicity. There are, however, losses during turn-off proportional to the switching frequency.2

Another control scheme, based on the boost converter in the CCM operation, replaces the fixed output reference with one proportional to the input voltage level. This is called a boost-follower circuit. For continuous input current operation in a boost converter, the output voltage must remain higher than the peak input voltage. This is achieved in the boost-follower arrangement but with the benefit of a reduced switching frequency at low line, which translates into a reduced turnoff switching loss.3

Yet another scheme operates the switch at a fixed frequency but varies the duty cycle. “To meet EN 61000-3-2 current harmonic limits, the Model 61504 AC Source has adopted a boost PFC circuit using the Texas Instruments UC3854A Preregulator as a controller,” said Galen Chou, product manager at Chroma ATE. “Using average current- mode control and fixed-frequency switching, the system is inherently stable under almost all line and load conditions. Even with PF >0.98 and full load, current harmonic distortion is constrained within a very small range.”

Xantrex also uses a fixed-frequency boost circuit, according to Mark Edmunds, the company’s director of engineering. But for units operating from high-voltage AC input lines, such as 380 VAC, a fixed-frequency buck configuration is used, which allows the internal DC link voltage to be kept to about 400 V.

To View AC Sources Chart Click Here

To View Features2 Chart Click Here

Improved PFC Efficiency Because PFC often is a separate section added in series to a power supply, overall efficiency will be lowered. For that reason, many circuits have been proposed to improve efficiency. At the expense of increased complexity, zero-voltage-transition (ZVT) switching can be implemented that reduces losses in both the main switch and the rectifier diode. Figure 3 shows a boost converter modified by the addition of ZVT components.4

The added capacitor Cr and inductor Lr form a resonant circuit controlled by the switch Qa. In practice, sufficient output capacitance already may exist, or a separate capacitor may be added.

ZVT circuitry reduces switching losses in the main switch by driving the voltage across the switch to zero before the switch is turned on. The resonant circuit discharges the switch drain-source capacitance so the stored energy is not dissipated when the switch is turned on. The circuit also limits the rate of current change in the rectifying diode as it turns off. Because the charge stored in the diode is related to the rate of current change, less stored charge results. This, in turn, means that the reverse recovery current is lower, and diode turn-off losses are reduced.

Rex Reed, a Kikusui area sales manager, said that the PAS Series of DC supplies sold through Aeroflex used boost PFC circuitry operating in a continuous current mode. To further reduce noise and improve efficiency, the supplies may use an edge-resonance type of circuit in the future. And, Saul Kupferberg, Kepco’s vice president of sales, commented that the company’s supplies used a resonant form of average PFC (APFC) that resulted in substantial power savings when compared to conventional hard-switched circuits. ZVT has the benefit of affecting operation of the basic circuit only during the commutation intervals, when the switch and rectifying diode are changing state. Other schemes force the current to zero prior to switching. Resonant and quasiresonant converters use this technique, but their operation differs fundamentally from that of the basic boost PFC circuit.

Summary

An increase in the prevalence of PFC is only one of the trends you will find in new power supplies and sources. According to Gerald Holland, director of sales for California Instruments, you also can expect high-power sources to have more control and analysis capabilities. In addition to the usual IEEE 488.2 and RS-232 interfaces, Ethernet is becoming more available. And, watch for new European regulations that may affect your power application. More generally, Allison Harvey, the wireless program manager for power products at Agilent, sees a continuation in requirements for lower voltages and higher peak currents. Look for supplies with greater accuracy and stability because there is less room for error when operating at low voltages.

Kikusui’s Mr. Reed also emphasized low-voltage/high-current requirements. In addition, he commented that fast transient load response was being driven by high-speed microprocessors. He also referred to the beneficial environmental effect of very high-efficiency supplies and those that can return load dump energy to the input power utility.

Power supplies are commonly integrated as part of a test system, and in these cases, digital communication to a PC is important. Jeremy Simon, director of sales at Glassman High Voltage, said that analog interfaces continued to be standard on its products, but RS-232, IEEE 488.2, and Ethernet were gaining in popularity among users.

An important test-system concern is throughput. Common to both Elgar’s DLM 600W Series and Agilent’s N6700 Modular Power System is a feature called down programming.

There are many more aspects of today’s supplies to consider, such as front-to-back cooling that really does allow you to put other instruments immediately above and below a 1U power supply. But, despite the large number of products, buying a DC power supply or AC source is a matter of finding the best match to your budget and application. The excuse can hardly be made, given the models available today, that your technical requirements cannot be satisfied.

References
1. Grigore, V., Topological Issues in
Single-Phase Power Factor Correction:
Dissertation for the Degree of
Doctor of Science in Technology,
Helsinki University of Technology,
Nov. 30, 2001, Institute of Intelligent
Power Electronics Publications, Publication
6, http://lib.hut.fi/Diss/2001/
isbn9512257351/isbn9512257351.pdf
2. Turchi, J., Power Factor Correction
Stages Operating in Critical Conduction
Mode, ON Semiconductor,
AND8123/D, September 2003, Revision
1, http://www.onsemi.com/pub/
Collateral/AND8123-D.PDF
3. O’Loughlin, M., Advantages Using
a Boost-Follower in a Power Factor
Corrected (PFC) Pre-Regulator, Texas
Instruments, July 2002, http://
www.analogzone.com/pwrt0708.htm
4. Sasic, B., Design-Oriented Analysis
of the Two-Output Isolated Soft-
Switching PFC Circuit, Kepco, 1996,
http://www.kepcopower.com/
bpaper.htm

April 2004