TI Training home
Power Supply Design training resources
Power Supply Design Seminar 2018 technical resources
Power solutions for Class-D audio amplifiers Power Solutions for Class-D Audio Amplifiers in Automotive Applications
Power Supply Design training resources
4.4 Power Solutions for Class-D Audio Amplifiers in Automotive Applications
Part 4 covers an automotive power supply solution. In automotive high fidelity or high fi, the maximum power capabilities of an amplifier play a very important role because the parameter's often solely used for comparison of different amplifiers and systems. This is not limited to only aftermarket amplifiers. Car manufacturers also emphasize having large numbers in their order specifications. It's arguable that the declaration of peak RMS power without the corresponding THD makes any sense, but this is typically a relationship that doesn't play a big role.
For this example, a customer requested a power supply for a 700 watt audio amplifier. First we will take a detailed look into the power specifications of a TPA3251 class D amplifier to understand the different output power specifications and which power supply demands arise from that.
The amplifier system contains one power supply and two TPA3251 amps. Amplifier one is configured in bridge-tied load or BTL mode, and provides the full audio spectrum to the left and right channel with 4 ohm speakers. Amplifier two is configured in parallel bridge-tide load mode, or PBTL, and contains an 80 hertz low-pass filter on the input to drive a subwoofer with 2 ohm impedance. It is also possible to connect two 4 ohm subwoofers in parallel to gain the same amount of power.
When reading the specification of a class D amplifier, it's very important to understand the relationship between RMS power, peak power, and THD. At the beginning of this presentation, we stated that 0.1% THD is a value most people will not be able to hear. For this amplifier in bridge tide load mode, you can achieve a maximum power of two times 120 watts RMS and 2 times 240 watts peak with 0.1% THD for the left and right channel and 4 ohm speakers.
For the same THD and a single 2 ohm speaker, the total power for the subwoofer channel is identical. If you increase the THD to 10%, the RMS and peak levels are around 46% higher. So it's always very important to check if an amplifier is specified either with RMS or peak power and at which THD.
In this example, the complete system can be specified either with 480 watts taking RMS power and 0.1% THD into account, or with 1,400 watts taking peak power and 10% THD into account. The difference is almost three times.
The prestage of this audio system is designed to provide full power at 10% THD with an input level of 0 dBu. 0 dBu is a standard for European pro audio applications and equals 2.2 volts peak to peak at 600 ohms. The input power source for the amplifier is the car battery and must provide full power between 9 and 16 volts, and must be operable down to 6 volts with reduced power.
For the design of the power supply, 0.1% THD is then taken into account, which equals 380 watts RMS and 960 watts of peak audio power. Considering a crest factor of 1 to 8 for this kind of application results in 120 watts of RMS continuous power. Since this is the audio power directly on the speakers, the efficiency of a class D amplifier and the power supply itself have to be taken into account. Assuming 90% efficiency for the amplifiers and 95% efficiency for the power supply, the output current of the power supply is 3.9 amps average and 30.2 peak at 36 volts.
It is unlikely that both amplifiers need the maximum power at the same time. Therefore, the current limit of the power supply is limited to 20 amps. This is a current at which all components of the power stage must be able to remain in operation for a short time, in the range of a few seconds. The power supply and its power stage, including the thermal interface, need to be thermally designed to provide 3.9 amps continuously without any limitations. The most reasonable power supply topology for this application and spec is an interleaved boost converter as no galvanic isolation is needed and it supports high efficiency over a wide load range.
Two LM25 122 synchronous boost controllers are used in master-slave configuration and are switching with a 180 degrees phase shift to reduce the ripple current stress on the capacitors. The audio signal is preprocessed by OPA16628 low noise and distortion audio op amps. The class D amplifier for driving the subwoofers has an additional active low-pass filter with a cutoff frequency of 80 hertz.
The component selection of the boost converters is driven by several factors, but the main one is the peak current demand. That is also the reason for using the two-phase approach instead of the single-phase boost converter, which should be sufficient to provide an average current of 3.9 amps continuously.
For the high-peak current, which is limited to 20 amps, the low-side FETs are put in parallel to enable peak power capabilities and to spread the heat. As the output current is relatively low compared to the input current of the boost, the same effect for the high side can be used.
This design uses flat wire inductors with ferrite cores to maximize the efficiency while providing a very high saturation current. A more cost-sensitive approach would be to use inductors based on iron powder, which reduces the mechanical footprint.
At the output of a boost converter, the current is pulsed as energy is only delivered during the off time. Additionally, the class D amplifiers which are connected to the output of the boost converter look like a synchronous buck converter operating at a 50% duty cycle. In a buck converter, the stress is opposite. There's large AC current stresses on the input and low AC current stresses on the output.
Finally, the boost output capacitors have to cover not only the AC current stress from the power stage itself but also the pulse input current of the amplifiers. To reduces this stress, a post-filters is used on the output of the boost converter which not only reduces the current stress for the electrolytic capacitors but also provides a very clean supply voltage for the amplifiers.
Directly connected to the synchronous rectifiers are eight parallel 4.7 mic ceramic capacitors which are capable of handling any currents in the range of several amps per capacitor. The disadvantage is the load capacitance, which results in a voltage ripple around 570 millivolts peak at 13.8 volts input voltage and 10 amps load.
A small inductance of 100 nanohenrys after the ceramic capacitors of a boost, together with the electrolytic capacitors of a class D amplifier, form a low-pass filter which provides a clean supply voltage for the amplifiers and decouples the boost AC current from the amplifier's input capacitors. The cutoff frequency of a low-pass filter is set to approximately 1/10 of the switching frequency, which theoretically results in 40 dB of ripple rejection. 25 kilohertz is also high enough to avoid any negative impact on the converter's bandwidth.
For post-filters with a high quality factor, a damping resistor in parallel to the inductor might be necessary to avoid resonance. The inductor used has a DCR of only a few milliohms. But together with the relatively low ESR of the electrolytic capacitors, the simulation showed no resonance and that the damping resistor was not actually necessary.
The left picture shows the efficiency of a boost converter at 36 volts on the output and 10 amps load. The measurement shows an almost flat curve over a wide load range with a peak efficiency of 95.5% at 13.8 volts input voltage.
At 9 volts input, the efficiency is around 3% lower, caused by increasing conduction losses due to the higher input current. The picture of the converter's bandwidth on the right side at 10 amps load doesn't show any surprises and is a bit below 1 kilohertz. Electrolytic capacitors can have significant changes in ESR over temperature and lifetime. Therefore the compensation was designed to provide a generous phasing gain margin of 81 degrees and negative 21 dB, respectively.
The system audio performance is excellent and the THD is well below 0.1% over the whole frequency range from 20 hertz up to 20 kilohertz for 12 volts input voltage. The left picture shows the measurement results for 25, 50, 75, and 100 watts RMS at a 4 ohm load. The THD versus power measurement on the right picture shows that up to 100 watts RMS, the THD of the system is below 0.1%. This fits exactly with the specification of the TPA3251 class D amplifier data sheet, which guarantees 0.1% THD at 120 watts RMS for a BTL or bridge-tied low channel at 4 ohms load.
The measurement proves that the power supply has no negative impact on the audio performance and is comparable with the lab reference supply which was used to specify the class D amplifier IC. This slide shows both the size for the AC to DC power supply on the left side and the automotive audio system on the right side. All documentation for these designs is available on ti.com.
Texas Instruments class D audio amplifiers have the ability to deliver clean signals even when the supply voltage is not that clean. Minimum output filter capability has been calculated keeping audio performance degradation in mind. An AC to DC power supply with DFC and an automotive solution have been presented. Test results for both cases show that there is almost no difference between the reference power supply and our solutions. Further analysis might include efficiency improvements over different DC to DC topologies like the half bridge versus the two switchboard versus the LLC resonant topologies. Thank you for listening.
February 8, 2018
This training series describes how to properly design a power supply for a high-power Class-D amplifier based on the output impedance requirements as well as on typical requirements like average and peak power demands. In this video, we explore a high-power automotive audio amplifier solution.
This course is also a part of the following series