Paul McGowan writes:
The last thing you want to do to a power amplifier is put something with impedance between it and the loudspeaker. It’s one of the reasons we spend so much time and money making sure the speaker cables we use are as thick as a horse leg, impede nothing and connect our amplifier output with our loudspeakers as if the amp were sitting an inch away from the speaker terminals.
Tube amps and a few solid state amps use output transformers to couple their output to the speaker and this causes a lot of sonic issues and defeats the idea of making sure there’s nothing impeding the path. For example, as the loudspeaker impedance changes up and down with the music, a transformer can’t keep up and you wind up losing loudness at certain frequencies – the exact opposite of what we want – and this causes the tonal balance to shift in a most unwelcome manner.
And so it is with Class D amps as well – only they don’t have an actual transformer on their output – they require an output filter that is half a transformer plus a few other parts. This filter is directly in the path of the music and stands between the amp output and the loudspeaker input.
Why do they need this filter?
To take out the unwanted switching noise – the garbage – noise we do not want to get into our speaker cables or our loudspeakers.
Up until just a few years ago this situation created a lot of sonic problems – in the same way that output transformers are problematic. Most Class D amps performed differently into different impedances so, if your loudspeaker was a 4 Ohm or an 8 Ohm device, the amp would sound differently into each of those speakers – not to mention the problem that no speaker has a constant input impedance.
Our first Class D amp, the HCA-2, performed differently into changing loads and we did what we could with the technology of the day to smooth this effect out as much as possible. Much has been learned and developed over the last decade.
The best Class D amp technology today has solved these problems by using global feedback around this filter – eliminating any differences the filter may contribute – and the output filter problem has simply gone away. To engineer this requires some extremely sophisticated computer modeling techniques and an extraordinary level of skill in what many term “black magic” – so difficult is the mathematics and understanding – of Class D.
Few understand this as well as Hypex engineer Bruno Putzey who managed to not only get the output impedance of the class D amplifier down to micro Ohms at any frequency, but he’s done so without sacrificing any performance or stability – something you’ll just have to take my word for with respect to the level of difficulty involved. There are other low impedance designs out there but none I am aware of that perform anywhere close to the Hypex.