Efficiency and interference of Class D audio amplifier in use

From the previous discussion on "How to evaluate the dynamic efficiency of a Class D audio amplifier", if Poq represents the ratio of output power Po to static power consumption Pq, Emos represents the efficiency of the output power transistor and Eff represents the total efficiency, then the total efficiency is:

Eff = （ Poq x Emos ） / （ Poq + Emos ）

The above formula shows that the influence of Poq on the total efficiency is the same as the influence of the efficiency of the output power transistor Emos on the total efficiency, so the static power consumption Pq or the static current consumption Iq needs to be taken into consideration when evaluating the total efficiency.

Figure 1

Figure 1 shows the Eff curves of two stereo Class D amplifiers. The power supply of these two amplifiers is 12V and the efficiency Emos of the output transistors is 0.8. The quiescent current of the amplifier in curve A is 10mA and the quiescent current of the amplifier in curve B is 25mA. Because it is stereo, the quiescent current of each single channel is only half, that is, the quiescent current of curve A is 5mA and that of curve B is 12.5mA. It can be seen from the figure that the total efficiency of curve A is significantly better than that of curve B at low output power or when the output power is about 10% of the maximum output power. Curve A is the Eff curve of TMPA430DS at 12V and 8ohm resistance load.

Figure 2

Figure 2 shows the Eff curves of two stereo Class D amplifiers. The power supply of these two amplifiers is 5V and the efficiency of the output transistors is 0.9. The quiescent current of the amplifier in curve C is 5.6mA and the quiescent current of the amplifier in curve B is 10mA. Therefore, the quiescent current per channel of curve C is 2.8mA and that of curve D is 5mA. From the graph, we can see that the advantages and disadvantages of the two curves are equivalent to those in Figure 1. Low quiescent current has high total efficiency Eff. Curve C is the Eff curve of TMPA3155DS at 5V and 8ohm resistor load.

Figure 3

Figure 3 also shows the Eff curves of two stereo Class D amplifiers. Curve F and Curve D in Figure 2 are the same amplifier, and Curve E and Curve C in Figure 2 are the same except that the efficiency of the transistor is only 0.85. Figure 3 shows that although the efficiency of the transistor in Curve E is only 0.85, which is lower than Curve F's 0.9, the quiescent current of Curve E is 2.8mA, which is lower than Curve F's 5mA. Therefore, when the output power is less than 0.18W, the total efficiency of Curve E is higher, which emphasizes the impact of low quiescent current on the total efficiency Eff.

When evaluating the total efficiency, it is not necessary to consider the switching loss of the output transistor when switching. In order to avoid short current or through current when the output power transistor is switched, a dead zone is reserved when the output power transistor is switched. Therefore, the corresponding NMOS of any output terminal must be completely turned off before the PMOS is turned on, and vice versa. Therefore, basically, the output NMOS transistor does not cause switching loss. However, the output PMOS transistor will charge the capacitance presented to the output terminal during the conduction period. This capacitance includes the bonding pad capacitance in the IC, the drain capacitance of the output transistor, and other stray capacitance. This charged capacitance will discharge when the NMOS is turned on, and the power consumed is already counted in the static current or static loss. If the two ends of the BTL are fully symmetrical, the static current will not increase when the load is connected. If the static current increases, it means that the load has a capacitive component.

Due to efficiency, when the power amplifier drives the speaker, it will generate heat and cause the temperature to rise. The temperature rise will cause the output transistor conduction efficiency Emos to deteriorate and the overall efficiency to decrease. Therefore, proper heat dissipation is necessary for high-power or low-efficiency power amplifiers. If the voltage gain of the Class D power amplifier is determined by the pre-circuit rather than the output end feedback, when the temperature rises, the output signal will shrink as the output transistor conduction efficiency Emos deteriorates. The reduction is determined by the output transistor conduction efficiency Emos and the size of the load. Since the on-resistance of the output transistor increases, the output current decreases, and the current of the power supply also decreases. If the voltage gain of the Class D power amplifier is fed back by the output end, when the temperature rises, the output transistor conduction efficiency Emos deteriorates. In order to maintain the voltage gain, the power amplifier must maintain the size of the output signal, so more heat is generated. Of course, the current of the power supply also increases.

In order to reduce electromagnetic interference (EMI), a magnetic bead (BEAD) filter is required at the output end. The filter capacitor behind the magnetic bead is usually 1nF. The output PWM signal charges and discharges the capacitor once per cycle. For a 5V power supply and a 250KHz operating frequency, the static current generated by each capacitor is

I = CVF = 1nF x 5V x 250K = 1.25mA

The BTL output stereo amplifier has 4 output terminals or 4 output filter capacitors, so the quiescent current increases by 5mA after adding EMI filtering. From the above formula, it can be seen that the increase in quiescent current can be reduced by choosing an amplifier with a lower operating frequency or a smaller filter capacitor.

Class AB audio amplifiers are also highly efficient at high output power, with an actual maximum efficiency of about 65%. Therefore, even if the efficiency of a Class D audio amplifier can be as high as 90%, the power it saves at high power output is only 25% of the total power consumption compared to a Class AB amplifier. Even so, the two efficiencies are very different in heat dissipation. If both amplifiers consume 20W of energy, the Class D amplifier outputs 18W of power and generates 2W of heat, but the Class AB amplifier outputs 13W of power and generates 7W of heat. If the Class AB amplifier outputs 18W of power, the heat generated will be as high as 9.6W. 2W of heat dissipation can be achieved using a general and inexpensive package, but 9.6W of heat will consume heat dissipation costs and space. In use, audio power amplifiers are not always used at maximum power. For a Class D audio amplifier with a maximum output power of 3W, if the output power is 0.5W or 1W, the efficiency calculated by the formula is already higher than 86% or 88%. For a Class AB audio amplifier with a maximum output power of 3W, if the output power is the same at 0.5W or 1W, its efficiency is only 25% and 37%. Therefore, the efficiency of a Class D amplifier is about 3 times that of a Class AB amplifier. The lower the output power, the greater the difference in efficiency. The efficiency of a Class D amplifier is much better than that of a Class AB amplifier, and it is more obvious when the output power is below half the power. If the audio content is a music signal, the output power is low most of the time, and the efficiency of a Class D amplifier is much better than that of a Class AB amplifier. If the audio content is intermittent, such as a news broadcast, the intermittent time is relatively long. During the intermittent period, only static power is consumed. For an amplifier with the same 3W output, the static power consumption of a Class D amplifier is only 10% - 20% of that of a Class AB amplifier. Experimental data shows that when a Class D amplifier uses a battery to play music, the battery life is more than 5 times longer than that of a Class AB amplifier.

The biggest disadvantage of using a Class D amplifier is that it generates interference signals. Since the output signal of a Class D amplifier is a pulse with a large current and fast switching, the interference signal is strong and the harmonic frequency of the interference is wide, which can easily cause poor reception of the receiver. The interference mainly comes from wiring conduction or radiation. To avoid wiring conduction, you can connect magnetic beads and filter capacitors in series to filter high-frequency harmonics. As for low-frequency harmonic interference, you can use an LC output filter, but the parasitic capacitance of the inductor should be small to avoid high-frequency interference from passing through the inductor. If the receiver and the amplifier are placed on the same PC board, the power supply and grounding should be isolated. It is best to separate or supply power separately at the power supply end. As for radiation interference, the antenna end of the receiver and the output end or wiring of the amplifier are placed on both sides of the PC board to increase the distance. The direction of the antenna should be perpendicular to the output signal running line of the Class D amplifier to minimize the antenna's reception efficiency. The shorter and thicker the lead or speaker wire of the amplifier output pin on the PC board, the better, to reduce the antenna effect, that is, reduce the antenna radiation efficiency and reduce electromagnetic radiation. If the design or stereo requires the use of long speaker cables, the most effective way to prevent radiation is to use speaker cables with isolation wires. The isolation wires are grounded at one end of the PC board and left hanging at the other end of the speaker. Experiments have shown that this shielding method has a considerable improvement when used in TMPA3155/3156. In addition, the proper placement of larger metal components can also shield some radiated signals. Since the interference signal comes from the switching of PWM, a lower operating frequency or PWM frequency can linearly reduce the interference energy. The operating frequency of Zhenyi Technology's Class D amplifiers is mostly set at 250KHz except for special considerations. In addition to EMI considerations, it can also reduce the increase in static current caused by the EMI filter capacitor mentioned earlier.