Quasi-Class E Achieves Power Control and ZIS/ZVS


David Pacholok, Paul Reich and Jim Spangler

This is the first in a three-part series describing a novel approach to high frequency (27 MHz) power conversion and control. This article describes the basic Quasi Class E circuit. Part 2 will describe voltage controlled reactance means to vary power. Lastly, part 3 will describe rectification to DC using SiC diodes and EMI issues.

 

Achieving Power Control with Varied Frequency

To meet the growing need for increasingly smaller power supplies, wireless chargers, and other devices, resonant power converters are being driven to progressively higher frequencies.

Off-line operation of MHz power converters presents unique challenges with regulatory agency EMC standards. Choosing a frequency of operation at random may lead to severe limits for conducted and radiated EMI, as mandated by FCC Part 15 rules, for example. Operating the converter at one of Part 18 industrial/scientific/medical frequencies may simplify EMC compliance.

However, fixed frequency operation reduces design freedom in terms of topology. For example, the popular LLC converter is controlled by altering the switching frequency to change output power/voltage. Fixed frequency operation is not an option for the LLC and other topologies.

About 40 years ago Nathan Sokal taught the world his class E RF power amplifier technology 1,2. This topology, while very efficient and cost-effective, has one major issue for HF switch-mode power supply application, or any application where power/voltage output needs to be varied. This requirement covers perhaps 90% of possible applications. In order to vary the output, whether rectified-to-D.C. or simply RF, the D.C. input voltage to a class E amplifier must be varied. Hence, a pre-converter stage of some form is needed, such as a buck converter. For some applications the size, cost, and efficiency penalties of the pre-converter are tolerable, but for cost-sensitive consumer applications, they are not desirable.

We have developed a GaN-based quasi-class E topology that allows power output control over a 10:1 range while maintaining efficient commutation. This may be defined as Zero Current Switching (ZIS) at device turn-on and Zero Voltage Switching (ZVS) at turn-off. This opens the door for high-efficiency DC-DC converter applications. As an RF generator, besides variable output, the advantage of operation over a wide range of load impedances is provided while maintaining efficient commutation.

Our concept is to vary the reactance of a control element to vary power. A variable reactance tunes a series or parallels reactive network up and down its resonance curve to control power output.

The variable reactance element may take the form of a voltage variable capacitance, a saturable reactor, or a digitally switched array of capacitors or inductors as we will show in Part 2 of this series.

In this article, we simply vary the value of C2, as shown in Table 1, to achieve power control.

 

The Basic Resonant Flyback Circuit

Referring to the resonant flyback converter topology of Figure 1, L1 and C1 (including Cds) provide a resonant frequency higher than the drive frequency. The half-sine drain voltage waveform returns naturally to zero before the next “on” time. With very light loading, the inductor current tends toward a sawtooth (Figure 2).

 

Basic Resonant Flyback Circuit

Figure 1: Basic Resonant Flyback Circuit.

 

Simulated Drain Voltage

Figure 2: Simulated Drain Voltage (green) and Inductor Current (red) (Figure 1 Circuit).

 

C1 and L1 were chosen to have a characteristic impedance (ZO) of about 100 ohms:

 

(1)

ZO = (L/C)1/2

 

This value reduces losses in the GaN FET, L1, and C1.

 

Evolution of the Quasi-Class E

We started with the classic class-E equations 1. Our operating point was set to 100 watts output with 100 VDC input, at the 27.12 MHz ISM Frequency. While the topology of both the class E and our quasi-class E (Figure 3) are identical, the C1/C2/L1/L2 values in Table 1 are very different.

 

Comparison of “Classic” and Proposed Components

Table 1: Comparison of “Classic” and Proposed Components.

 

Generic Class E Topology

Figure 3: Generic Class E Topology.

 

The 24 ohm load requirement derived from the Sokal equations 1,2 did not match our available supply voltage, desired load impedance, or required power output. 50 ohm loads and power meters are common; 24 ohm units, not so much.

 

Minimum Power Simulation

Figure 4: Minimum Power Simulation.

 

Operation in the inductive portion of the L2-C2 series resonance curve is mandatory. If the tank circuit resonant frequency exceeds the drive frequency, the network becomes capacitive and ZVS and/or ZIS will be lost. Poor efficiency, poor commutation, and possible device destruction will result.

The inductive load causes the drain of the switching device to be driven a few volts below ground at the beginning of the “on” time. The gate is relatively positive, turning the device slightly on. This causes a small V*I loss. This loss may be reduced by turning the switch on at the moment of Vds=0. Practically speaking, square wave drive at 27 MHz is difficult. Sine wave drive from a resonant gate drive network can provide gate drive energy recovery. In the case of sine wave drive, the use of positive gate bias at or slightly above device threshold will turn on the FET earlier, reducing its I*R drop during reverse current. GaN FETs have no body diode, so charge storage effects do not occur.

 

Maximum Power Simulation

Figure 5: Maximum Power Simulation.

 

To control the output power, a series network consisting of a fixed 1.3 uH inductor (L2) and a variable capacitor (C2), shunted with a fixed capacitor (to limit the capacitance range) was added. The output power into a 50 ohm load was measured after considerable simulation and adjusting C1 and L1. Maximum output power with good commutation was noted for C2 values at or slightly above 30 pf (L2 and C2 were near resonance). Increasing C2 dropped FR and moved the network away from resonance. This controlled power output over a 10:1 range is shown in Table 2.

 

Power Output vs. C2

Table 2: Power Output vs. C2.

 

The loaded Q of this network was 4.5 with a 50 ohm load.

 

(2)

Q=XL/R

(3)

FR = 1/(2π (LC)1/2)

(for L2= 1.3 uH, C2 = 29.5 pF, FR is 25.7 MHz)

 

The need for an inductive load line and the resulting reverse FET current are major departures from conventional class E design. We noted less than 1% efficiency reduction with our topology at maximum power, compared to a similar class E circuit with constant supply voltage and output power.

 

In the next article of this three-part series, we will investigate various means of reactive power control.

 

References

  1. N.O. Sokal, WA1HQC, “Class E RF Power Amplifiers”, QEX Magazine, Jan/Feb. 2001, ppg 9 – 20
  2. N. O. Sokal et al, “High Efficiency Tuned Switching Power Amplifier”, US Patent 3919656, Nov 11, 1975

 

About the Author

David Pacholok has operated Creative Electronic Consultants LLC since 1981. He is active in switchmode power conversion, electronic lighting ballasts, RF power, pulse Power, and induction heating. 

Paul Reich is a freelance consultant. His specialties are radio frequency design and simulation. In a previous life, he held various positions at Motorola, Inc. and the Chamberlain Group. Paul earned his BSEE in 1976.

Jim Spangler is President of Spangler Prototype Inc. Jim was a Field Applications Engineer at Motorola Semiconductor, On Semiconductor, Cirrus Logic, and Active Semiconductor. Jim was involved with several IEEE societies and planning for the PSMA Applied Power Electronics Conference (APEC). Jim has published extensively about power electronics applications. 

More information:    Source: Bodo's Power Systems, November 2018