Class D amplifiers have increased their penetration in the audio world as a result of their efficiency, size, and solution cost advantages. Moreover, with “green” initiatives now in place, the 30% to 40% reduction in power loss delivered by Class D solutions in comparison to Class AB amplifiers is attracting widespread attention.
However, there are some intricate topology-related drawbacks associated with Class D amplifiers. One such drawback is the increased distortion that Class D IC amplifier designers need to overcome.
A marked cause of degradation in Class D amplifiers (Fig. 1) is bus pumping, which can be seen when the half-bridge topology is delivering low-frequency output power (subwoofer range) to the load. The gain of a Class D amplifier stage is directly proportional to the input bus voltage. Bus fluctuations resulting from load current, therefore, create a second-order variation of gain, causing distortion.
Because the energy flowing in the Class D switching stage is bi-directional, there is a period where the Class D amplifier feeds energy back to the power supply. Most of the energy flowing back to the supply comes from the energy stored in the inductor in the output low-pass filter.
Usually, the power supply has no way to absorb the energy coming back from the load. Consequently, the bus voltage is “pumped up,” creating bus voltage fluctuations. Bus pumping does not occur in full-bridge topologies because the energy kicked back to the power supply from one side of the switching leg will be used in the other side of the switching leg.
In a high-performance Class D amplifier in half-bridge mode, the capacitors across the +B and –B supplies (Fig. 2) (right side of the schematic) will have to absorb the pumped energy during low-frequency bus-pumping cycles.
Synchronous Bus Converter Power Supply
The input bus power-delivery scheme will impact the bus-pumping phenomenon. Consider a synchronous secondary switch-mode power supply (Fig. 3).
The bus converter is a 50% duty cycle converter (actually about 49%, due to the programmable dead time inserted to prevent cross conduction). The secondary output stage performs as a current doubler, allowing current to be delivered to the load via the synchronous MOSFETs, 98% of the time. In the positive transformer secondary winding, node P1 and the gate of Q2 are in phase. The same is true of their complements (Fig. 4).
Forward current flows during each half cycle in either Q1 or Q2 (green and blue lines), delivering power to the load. As a “buck-derived” topology, the circuit operation is similar to a two-phase buck converter, at 50% duty cycle, i.e., current is sourced (and sunk) 100% of the time (except of course for the dead-time previously mentioned). While it is sourcing current, the synchronous converter behaves the same as an asynchronous converter (diodes instead of FETs), except with higher efficiency.
Continue on next page
If the converter were operating asynchronously, the commutating current flowing from the Class D amp (dashed red lines) would be absorbed by the 1000-µF bus capacitor. Its impedance at 20 Hz, using Z = 1/(2π fc), would be 8 Ω.
This capacitor and the Class D output-filter inductor form an LC tank, in series with the load resistance. The cutoff frequency of this R-L-C tank determines the audio signal frequency below which bus pumping will be a problem.
To reduce the amplitude of the voltage that appears on the input supply, i.e., “pumped-bus voltage,” three remedies exist:
• Larger bus capacitors (the same energy will cause a smaller bus voltage increase)
• Higher minimum signal frequency; bus pumping is a problem for FAudio less than 200 to 500 Hz
• Lower power and/or higher load impedance (less energy transferred between supplies)
Synchronous Supply Sinking Impedance
The above analysis covers the “sourcing” case. The “sink” impedance of the synchronous output stage, on the other hand, consists of the FET on resistance added to inductor and circuit-board resistances. (In an asynchronous supply, it would be just the bus capacitance.) This is obviously much smaller in magnitude than when the supply is sourcing.
During the dead time, when neither Q1 nor Q2 is conducting, the circuit will behave like the asynchronous supply and will reflect the higher impedance. That, however, is only 1% of the time.
Oscilloscope screen captures demonstrate the advantage of powering a Class D audio amplifier with a synchronous converter (Fig. 5). The traces in Figure 5 (top) show an asynchronous supply. The traces in Figure 5 (bottom) show a synchronous supply.
Note that in the synchronous case, the ripple is at twice the frequency (40 Hz) due to the dual-phase nature of the bus converter while it is operating in sink mode. The amplitude of the asynchronous ripple is 9 V at 25 W per channel, while the amplitude of the synchronous-mode ripple is 2 V at three times the output power.
A further consideration is that, at 75 W, an asynchronous-mode power supply yields significant distortion if additional capacitors are not added to the dc buses. There is significant improvement in total harmonic distortion plus noise, or THD+N, (0.4457 versus 0.00685 or 36 db) using a synchronous power supply versus an asynchronous one at the maximum output power range. There is no significant improvement of total harmonic distortion (Fig. 6) with audio frequency higher than 200 Hz.
In addition to signal degradation, the increased ripple that’s on the input dc bus can cause an overvoltage to be applied to power devices, resulting in catastrophic failures. While products such as the IRS20955 Class D amplifier from International Rectifier have overvoltage protection and will shut down the switching amplifier if the input voltage (VIN) exceeds a user-programmable threshold, a synchronous supply will alleviate such concerns altogether.