This is an audio amplifier implemented on a PCB board. It is capable of driving audio signals with a frequency between 0-20kHz into a 4 ohms speaker using a 12V supply. The final design achieved 96.3% efficiency. This project was completed as a final project for my Power Electronics class as a group of two.

This system consists of a buck DC-DC converter that satisfies the specifications of greater than 95% efficiency, tracking reference up to 20kHz bandwidth, and voltage ripple output less than 100mV. To accomplish these goals, considerations were into designing the best inductor and capacitor combination for the powertrain filter to ensure tracking capability, as well as choosing the MOSFETs based on parasitic capacitance and inductor to achieve minimal power loss.

After completing the schematic design and simulations, we created a PCB layout for the schematics and later conducted soldering and testing.

How to maximize power efficiency?

The Buck DC-DC converter is a critical part of this system. Therefore, it is important for us to control the power loss to maximize the efficiency. Power loss comes from two major sources of loss: switching loss due to the switching of MOSFETs and conduction loss due to parasitic resistance. Less significant loss results from operations such as quiescent current drawn by the gate driver.

Therefore, to minimize power loss in our design, we calculated the worst-case total power loss (switching loss + conduction loss) of different switches available to us. The worst case is defined as duty cycle (D) = 50% — this is when we have the highest conduction loss (Isw, rms ∝ sqrt (D)) highest voltage ripple resulting.

For the results, we found that OptiMOS (TM)3 power transistor (IPD220N06L3) by Infineon has the lowest total power loss of 0.045W/switch at 50% duty cycle and a thermal resistance of 62K/W.

How do we drive the speaker?

The essential function of this audio amplifier is to take in input from the audio source and be able to drive this signal with high power through the speaker. This means the system will be able to preserve the audio signal and amp up its current to be able to drive a speaker (higher current, louder speaker).

To accomplish this, the converter architecture uses the buck converter to track the input signal and employs a low-pass filter to pass only the audio bandwidth of up to 20kHz bandwidth.

For this RLC circuit, there are four design considerations: a). bandwidth cut-off frequency> 20kHz b). Q = 1c). Voltage ripple <100mV d). |Vout/Vin| = 0dB at 10kHz.

After selecting capacitor and inductor values, as well as selecting the specific component we are using for the audio amplifier. We did theoretical calculations using LTSpice simulations before building the circuit. The simulation was done by simulating Vref as a 500mv offset sinusoid wave with Vamp = 0.129V (at 10kHz and D = 0.5, Vin = 12V). The output shows a sinusoid wave with a 5.8V offset and Vpp = 4V. This is close to the 6V offset we are expecting. The loss calculated was 0.1976W, which equals 97.5% efficiency.

Making the physical board: PCB Board Design and Soldering

We designed the board on Eagle PCB. It was designed to minimize the size of current loops so as to minimize the effect of unintended inductance. A ground plane was provided so the current reaching the ground could take the shortest path back to its source. Traces were laid out with no acute angles to help with current flows. After finishing the design, we sent our draft to get it manufactured. We later soldered all the components including banana jacks onto the board.

I am grateful for this opportunity to design and build this audio amplifier. There were a lot of difficulties along the way such as making the best PCB layout and making sure the components are soldered correctly and functioning. We were able to successfully capture the behavior of our board and prove that it was functioning. In this process, I learned that we should always feel very confident with our design and calculations before implementation. Not all problems will be exposed at this stage, but it will reveal some critical issues if they exist. Second, during testing, it is always good to start with low-voltage inputs and slowly ramp them up — our oscillator unfortunately broke during testing near the limits.