Hi
I keep seeing numbers between 250-450kHz being the "standard" for class D audio. However, I can't seem to find why this is actually the case. I figure that as the frequency ^ we get more PWM of the audio input, but why wouldn't I then just go for the highest frequency? Is it due to noise?
I read
Note: I'm using a comparator + integrator configuration to generate a triangle as the reference.
I keep seeing numbers between 250-450kHz being the "standard" for class D audio. However, I can't seem to find why this is actually the case. I figure that as the frequency ^ we get more PWM of the audio input, but why wouldn't I then just go for the highest frequency? Is it due to noise?
I read
Note: I'm using a comparator + integrator configuration to generate a triangle as the reference.
The ratio between the switching frequency and the maximum audio frequency is called "pulse number". A pulse number of 20 is often the recommended minimum for high-fidelity applications, placing the optimum switching frequency for 20kHz audio signals at about 400kHz.
Switching frequency is a compromize - too fast and the losses and non-linearities due to the switching transients(*) increase, too slow and its harder/more expensive to filter out the high frequency energy.
The technology of the switching device strongly affects the choice of frequency - for instance GaNFETs are much happier in the MHz range than silicon MOSFETs as they have considerably quicker switching speeds.
There's also the accuracy of ramp waveforms to be considered if its analog class-D - faster ramps are harder to make linear, and comparators perform worse and worse with steeper sawtooth waves.
(*) The nature of a MOSFET switching transient is affected by the current flowing in a complex and non very linear way.
The technology of the switching device strongly affects the choice of frequency - for instance GaNFETs are much happier in the MHz range than silicon MOSFETs as they have considerably quicker switching speeds.
There's also the accuracy of ramp waveforms to be considered if its analog class-D - faster ramps are harder to make linear, and comparators perform worse and worse with steeper sawtooth waves.
(*) The nature of a MOSFET switching transient is affected by the current flowing in a complex and non very linear way.
Member
Joined 2018
Hi ligidk-San,
Suppose you are interested in non-standard higher switching frequency amps. You can follow my TAS6422 design if you like.
Fully digitally connected 2.1MHz switching sound feels smooth and has more natural high-frequency tones than 400kHz switching amps.
https://github.com/CyberPit/TAS6422-AMP-Project
CyberPit
Suppose you are interested in non-standard higher switching frequency amps. You can follow my TAS6422 design if you like.
Fully digitally connected 2.1MHz switching sound feels smooth and has more natural high-frequency tones than 400kHz switching amps.
https://github.com/CyberPit/TAS6422-AMP-Project
CyberPit