Ha ha, yes indeed,

I forgot about the Twit Class.

Document uploaded.

The inductors and capacitors on the outputs are to stop the fast switching edges from radiating into the ether. They need to be able to handle high currents too. Yes, they are usually around the values you see in this design, higher powered amps need physically bigger components of course.

The first Class D design I built, about 12 years back, used a 20 watt per channel ROHM part. I fitted some "ordinary" small axial inductors while I was waiting for the proper ones to arrive, They went up in smoke as soon as I tried to feed more than half a watt into the load.

That design was very successful in production and one of them still powers my home setup.

Yes,

I imagine the component size of the LC network would be dictated by power.

With that in mind, and our comments about the general size of the TI amp, I speculated that the relatively large LC assemblies may be better fitted to the speakers, like the old speakers with transformers fitted onto them.

Yes BBTV, but that would defeat their purpose, the speaker wires would radiate.

If you are building a device with the AMP on the back of the speaker, that's a different story. Keep the total voice coil connection length under about 60mm and you don't need the filters. Any switching spectrum that gets to the voice coil is way above audible range and would be further attenuated into insignificance by the inductance of the voice coil and the inertia of the speaker.

Good point Ian,

Would shielded cable to the speakers ( or twisted) overcome that problem?

It might help, need to test it though.

It would be a problem to get such a device through EMC/EMI compliance testing.

So... the transistor unit could be used to replace the speaker transformer in an ordinary valve radio, and you'd get lots more audio output?

Sounds like an excellent idea.

Yes Rob, tongue firmly in cheek....

I wonder what 300 watts would do to a Rola 5F?