Class D output filter design

[thadman]

For Class D amplifiers, the linearity of the output filters is paramount with regards to sound quality.

What are the best type of output inductors? Aircore or ferromagnetic? Aircore will not suffer from saturation or core losses, but will have higher resistance for a given inductance and its magnetic field will not be contained. Will the necessary proximity of the inductor to the switching transistors negatively effect the performance of the circuit due to the generated magnetic field?

What are the best ferromagnetic materials and winding topologies?

What are the best type of output capacitors? It is obviously important that its properties do not vary dramatically with temperature or power. I have also heard of capacitors suffering from the piezoelectric effect, could anybody offer further insight on this? Multilayer ceramic capacitors (ex. those produced by AVX) look very interesting.

Thanks,
Thadman

[darkfenriz]

INDUCTOR:
air core is a bad idea because of high resistance and high capacitance, not to mention the size. Both gapped ferrites and distributed airgap powders have their pros and cons. Best powders are molybnedium permaloy (MPP) and slightly worse but way cheaper sendust materials (aka. koolmu, super mss, ...).

CAPACITOR:
basically SMD is a must, good quality film caps are good but rare, ceramic NP0/COG are the best choice, but are hard to find with higher voltage ratings. A bank of smaller value ceramic X7R is often the cheapest and still acceptable solution.

[Pafi]

The problem should be divided into two cases (at least): feedback after output filter and feedback before filter (or no feedback). Post-filter feedback makes filter linearity less important, but it needs a good high-freq behaviour.

[thadman]

Are any of the SMD inductors suitable? Vishay makes some smps inductors which can handle over 80 amps with very little resistance. They don't appear to be the pinnacle of linearity, but I'm not sure how much of a problem that would be if post filter feedback is used.

[darkfenriz]

They are probably drum inductors, which radiate a lot of EMI unless very well shielded.
E cores and toroids behave better.

[Pafi]

Vishay makes some smps inductors which can handle over 80 amps with very little resistance.

DC resistance is not really an important thing. Any value under 5 % of load impedance is just fine. Parallel losses, like hysteresis and eddy current are more important, since they act all the time even at idle (wich is the typical operation of an audio amp). Basically low DC resistance and low parallel loss are antagonists, so I wouldn't expect much from such a coil.

Are any of the SMD inductors suitable?

Mounting method is irrelevant.

Key parameters are:
- saturation current and characteristics (distortion),
- HF loss,
- capacitance,
- stray magnetic field,
- DC resistance,
- price,
- size,
- repeatability,
- mountability,
- etc...