improving efficiency with parallel output fets

[Davey T]

I was wondering if it would be possible to parallel and switch between different types of mosfets at the output of a class d to give maximum efficiency at both low and high output levels.

So at low volumes you would use a mosfet with a small die area giving the lowest switching losses but not so good rdson.

As the output demand goes up, you start switching on a different mosfet in parallel with a much lower rdson to take the current and reduce your i2r losses.

Has this been tried? Are there any problems which would result? How could you implement this?

-dave

[fredos]

Better to play with adaptive deadtime...Less deadtime at lower power, higger at full power, but with bigger fet, as usual with fast recovery diode...

[Eva]

If the amplifier is allowed to operate in resonant mode at low signal levels, there will be almost no switching losses. This requires some dead time that may result in increased THD for large signals.

There is no point in using several different types of switching MOSFET.

[Davey T]

thanks guys.

Can you link me to some threads or pdfs on amps which use resonant switching at low signal levels. Basically, my best current 10w VMCD amp pulls about 100mw quiescent power and i'm looking into ways of reducing this.

[Eva]

In this situation, "resonant" means that, during the high-to-low and low-to-high transitions on the switching node, the parasitic capacitances are charged only with the current flowing through the inductor and not by turning on the opposite switch.

All what is needed for this to happen is to allow enough dead time. By the way, 100mW idle losses is a quite good figure in my opinion. You have to account for gate drive power and for the power lost in the cores of the inductors.

[Davey T]

ah, i see. you were just refering to reducing shoot through current losses, not capacitance losses.

I've been looking into current mode class D used for RF amplifiers and the such. These resonate and give zero voltage switching of the mosfets. I couldn't find any attempts to build a current mode class d audio amp.

Yea, the TPA3001 i'm using don't have output filters and pull about 12mA at 11V although the datasheet says 8mA. This is by far the best i've found but i want to develop something better.

I find the Iq has the biggest influence on battery life.

[Eva]

The method that I discussed already provides zero voltage turn on. Half bridge class D stages are inherently resonant given enough dead time and inductance, resulting in no switching losses.