Crossover distortion in class D
Hi folks!
This is my first post at this forum. I was asked by Bjorn to write a few words about distortion in class D, in particular crossover distortion. I have not read the entire thread, so this might be a repetition of some issues mentioned earlier in the thread.
Crossover distortion is typically what happens when a push/pull amplifier changes state from one output device to the other. Class D amplifiers do have such changes too, but by definition, I'd say they do not really have crossover distortion. There are two reasons why I think it is important to distinguish between the crossover distortion phenomena and the type of distortion that is typically found in class D amplifiers.
1: Besides the fact that crossover distortion occurs when the amplifiers output devices changes from active to passive, this is also an important issue when investigating small signal performance, since crossover distortion also occurs when the signal crosses 0V, or in class AB amplfiers, close to 0V. In class D amplifiers, the output transistors changes state many times during a signal period even at the highest frequencies the amplfier can reproduce. The 0V crossing point of the signal has no effect on when the transistors cross 0V between them. This means in class D amplfiers we do not have a distortion phenomena that is linked to the signal crossing the 0V point.
2: One of the main issues in class D amplifier design is connected to the kinetic energy of the output coil. When the transistor pair at the output is at high voltage, the current runs from the transistors, through the coil, and out through the load and via the capacitor on the output. When the transistors change state, the coil also changes current direction. However, one of the main reasons for using a coil at all is its kinetic energy, or in other words, its ability to be slower than the switching frequency. After the voltage over the coil has been set up, there is a little delay before the current reaches its peak. This is when the open transistor closes. At this point there has to be a small safety delay before transistor no 2 opens. However, the coils nature of being slow makes it continue passing through current even though the two transistors completely disconnects the coil from the power source. When the current flows from the two transistors, through the coil, the voltage between the two transistors drops extremely fast. Without safety diodes it reaches thousands of volts in nanoseconds. The safety diodes shunts the current to one of the rails when it passes the rail voltage. This is what causes one of the most challenging distortions in class D amplifiers.
The nature of this distortion phenomena is that it is often depending on the signal level. In a well built class D, this type of distortion can be pretty much non existing at low signal levels. In a prototype reference design we have been working with, this sets in at around 25W. The distortion pattern is still low harmonic (mostly 2. and 3. harmonic) and at this point it is still below 0,007%. This design delivers about 100W at 0,1% THD, so one can say the main distortion components are pretty much non existing until 6dB below max power.
Unlike class AB amplifiers, this is a phenomena that gets lower or goes completely away at low levels.
So, I think that one can safely say that the type of crossover distortion you find in class AB amplifiers can not be found in a class D, and the class D does not have a similar crossover distortion phenomena.
. It is a good reference though.
