My chosen research topic for my BTech thesis was that of a Class D audio amplifier. This interested me as the thought of a relatively "cheap" and powerful amplifier appealed to me.
Figure 1: "Classic" amplifier topographies
The more "classic" topographies used with audio amplifiers are that of the Class-A, Class-B and Class-AB. These topographies can be observed in figure 1. These topographies change the potential at the load (speaker) through restricting or allowing current to flow to the load. This form of voltage amplification is extremely inefficient as current is flowing through the active devices while there is a potential across the device, causing significant power loss.
Figure 2: Block Diagram of a Class-D amplifierThe Class-D amplifier works very differently to that of the "classic" amplifiers. A basic function block diagram of the workings of a Class-D amplifier can be observed in figure 2. The input of the amplifier is put through a Pulse Width Modulator (PWM) or Pulse Density Modulator (PDM). This translates the input signal voltage to a series of pulses. The "D" in Class-D is commonly misunderstood to mean "Digital" as the output signal resembles somewhat that of a digital signal, but in reality both analogue and digital Class-D amplifiers exist. The classification of a Class-D being digital or analogue has more to do with how the PWM or PDM is achieved.
The average of the input signal applied to a PWM or PDM is proportional to the average of it's output. The PWM or PDM is sent through a gate driver and MOSFETs to increase the power capability of the PWM or PDM signal. As the MOSFETs are turned-on fully and off fully into and out of saturation there is usually a low potential voltage across the MOSFETs while current flows or a high potential voltage across it when little current flows. This makes for very little power losses and the majority of losses in a Class-D amplifier are switching losses not conduction losses as found in the "classic" amplifiers.
The increased signal outputted from the MOSFET bridge is then put through a low pass filter, this filter "demodulates" the signal, reconstructing a signal similar to that of the input, but with some amount of scaling.
I chose to design my prototype off an IRS2092 from International Rectifier, I am still in the testing stages, but my final amplifier with some luck will be 200Wrms, 22kohm input impedance, 300kHz switching and using IRF6785 DirectFet MOSFETs. I'm currently experimenting with a proof of concept amplifier at home, which is a 50Wrms design using IRF540N MOSFETs and 300kHz switching. See figures 3-7 for images of the proof of concept amplifier sections (50Wrms amplifier).
Figure 3: LPFs and volume control (bottom side)
Figure 4: LPFs and volume control (top side)
Figure 5: 80VA and 300VA transformer and simple power supply
Figure 6: Basic IRS2092 based amplifier PCB
Figure 7: Soldered Basic IRS2092 based amplifier
Figure 8: Top View of Basic IRS2092 based amplifier
Figure 5: 80VA and 300VA transformer and simple power supply
Figure 6: Basic IRS2092 based amplifier PCB
So far a lot of work still has to be done, more will follow once testing has commenced. :-)