class d amp design
There are many possible topology's for switching amplifiers
Crown audio's patented system seems to work well, as I understand it they have two 50% duty cycle square wave signals and they phase shift one carrier, instead of trying to directly generate very narrow pulses during low volume parts of the audio.
It seems to work better than amplifiers using the old triangle-wave-generator and compator approach.
**broken link removed**
Here is a really interesting document from an amplifier manufacturer called hypex, I don't remember enough about control theory to understand all of it.
**broken link removed**
>Class D amplifier - is a non-linear amplifier.
The overall transfer characteristic should be reasonably linear, it's just the internal switching that is on or off instead of linear
>Has very high efficiencies
From what I have seen 85% is about the most people manage in reality.
Loudspeakers can be significantly reactive at some frequencys, they don't look just like a resistor. Some amplifers have much better efficiencys than others when driving partly reactive loads.
>psrr = nearly 0dB
Overall PSRR of the whole amplifier can be ok with good design.
I'v seen a design with awful PSRR (worse than 0dB) because all
the analog signals were referenced to a middle-voltage generated
from the supply using two resistors.
>Mostly used as an audio amplifier
yes, though magnatec claim they sell lots (with response to DC)
for motor drives in guided missiles.
If this is for a high power amplifier I expect you will be trying to drive
the gates of big FET's from your ASIC?
Overload and short circuit protection in big amplifiers can be tricky to get right.