An excellent analogy for the comparison between Class-AB and Class-D amplifiers is that of a linear voltage regulator versus a switching voltage regulator. If the input and output voltage levels are similar, then the efficiency of both voltage regulator topologies is similar. The voltage-drop across the output transistor element in the linear voltage regulator is small, so the power lost is relatively small. However, if the input and output voltages are greatly different, the linear voltage regulator will operate with poor efficiency as the output transistor will have a large voltage drop across it. The switching voltage regulator will maintain a high efficiency since the output transistor will still be operating either in cut-off or saturation mode, transferring packets of energy from one voltage to the other.

High efficiency extends the run-time of battery-operated devices. In systems that are thermally limited, high efficiency reduces the need for a heat sink or, for the same heat sink, increased output power can be provided.

The efficiency of a Class-D amplifier is controlled by two components of loss -- switching losses, and dc losses due to the r_DS(on) of the output transistors. Each time the output transistors change state, there is some energy loss as the transistors are switching on and off. While attempts can be made to minimize the rise and fall times, this causes higher levels of electromagnetic interference (EMI) in the system due to the higher voltage slew rate. Also, faster rise and fall times cause increased losses due to the reverse-recovery characteristics of the other transistor's diode in that leg of the H-bridge. Of course, the trade-off is that the farther from an ideal square-wave output the system gets, the greater the distortion products due to the trapezoidal shape of the output voltage pulse.

It may be obvious, but switching the H-bridge at higher frequencies will also result in a degradation of the efficiency. Not only is there loss due to the switching losses (times the switching frequency), but there is also loss due to the gate charge that must be supplied and removed with every switching transition.

The efficiency of a Class-D amplifier, as Class-AB amplifiers, is also strongly dependent on the load resistance. Using a higher-supply voltage with a higher-impedance speaker to deliver the same output power, will result in a higher efficiency than using a lower-supply voltage with a lower-impedance speaker.

For the purposes of this discussion, it has been assumed that all the trade-offs have been made, and they result in an 80% efficient Class-D amplifier. Clearly, this value could increase or decrease based on the previous discussion.

For sine-wave signals at maximum output level, there is not much difference in efficiency between the Class-D (80%) and the Class-AB (67%). However, the significance of the Class-AB's low efficiencies at low-output power levels is more obvious when the crest factor is considered. Crest factor is a ratio of the peak-to-rms levels of a signal.

Click here for complete image.

Crest factors as high as 12 to 15 dB are possible in music. An amplifier capable of handling 6-W peaks will spend most of its time operating with an average output level of less than 200 mW. In this case, there is a great disparity between the efficiency of the Class-D and Class-AB amplifiers. Fig. 4 depicts a test pitting one of TI's Class-AB amplifiers (TPA0202) against TI's Class-D amplifier (TPA005D02.) The two amplifiers' output signals were made equal, and the same input music signal was fed to both amplifiers, with the audio source in a loop/repeat-mode. The supply used was a non-alkaline 9-V battery. The result (Fig. 4, again) depicts a nearly three-time improvement in battery life using the Class-D over the Class-AB amplifier!

For Further Study

Of course, there is no such thing as an engineering free lunch! Implementation of a Class-D amplifier is more challenging than a Class-AB amplifier. There are three areas that the designer must concentrate on: PCB layout, output filter design, and EMI/RFI. TI will soon be releasing information on these topics on our Class-D web page which can be accessed at: http://www.ti.com/sc/class_d.htm

Conclusions Amplifiers based on the Class-D topology are viable replacements for amplifiers based on the Class-AB topology, especially in equipment where efficiency is very important. Other applications for Class-D amplifiers are in thermally-limited equipment, where the Class-D's high efficiency allows the heat sink to be eliminated or, with the same heat sink, a higher output power can be obtained. However, there are several differences when using the two styles of amplifiers that the designer must take into account: PCB layout, demodulation-filter design, and the EMI generated by the switching signals.

About the author:

Don Dapkus is a Systems Engineer for Texas Instruments' Advanced Analog and Logic Products Group. He is responsible for TI's line of audio power amplifier ICs: new product definition, writing application notes, assisting customers with their design problems, and getting new products to market, both Class-AB amplifiers, as well as a growing list of Class-D amplifiers.