The Truth about Modern Class D
Yep, vaccinations cause autism, soy causes infertility and androgeny, the moon landing is a hoax, the Earth is flat, and... Class D is digital and invariably sounds terrible. No end to ever-popular conspiracy theories and urban legend, Ain’t it? But wait, has anyone seen where I mislaid my trusty high-end tinfoil hat? Aliens are after me again *Grins!* G. |
As it stands now the switching frequency noise output filter, has phase shift effects that reach right down to 5khz, that doubled for the worse at 10khz and doubled worse again 20khz. That's what many listeners find objectionable, in the upper mids and highs.This statement is false. The filter is set to be operational at a fairly high frequency, and are usually 12 db per octave filters. 6 db/octave filters exhibit phase shift over the widest range, so if the pole frequency is at 90KHz you could see some artifact at 9KHz but not at 5KHz. But most filters I've seen are 12db/octave and so don't have phase shift artifact anywhere within the audio passband. What many listeners likely 'find objectionable' is not the filter- its likely something else. Several issues exist that can account for that- higher ordered harmonics caused by poor loop feedback implementation, breakdown of the encoding scheme at higher volume levels, distortion from the input circuit... but its not the filter. IME a good class D amp has artifacts, but not the kind typical to traditional solid state, so in many cases I find especially the later generations to be more musical than traditional solid state. |
Unfortunately, this book is now out of print, but I think it would remind us of where we are : https://amzn.to/2M0bBLB |
@lukaske @toetapaudio Note the VERITAS Monoblocks from Merrill Audio used the Hypex NC1200 module for years. It was tested by many audiophiles and societies against other implementations and was found to still be the best. We now have a new proprietary Class D design from Merrill Audio on the Oganesson Line of amps. These are an order of magnitude better then the VERITAS or anything else in that group. What makes the difference in the use of GaN transistors. Tthe MOSFETS in the prior Class D designs, including the Hypex simply cannot do what the GaN transistors can. Other Class D designs, including the Hypex NC1200 utilize heavy feedback which tend to have loss of detail and become sluggy. Now imagine no feedback loops. That brings a new sense of ease and air to the music, along with a strong, fast attack, yet no overhang or edge. Soon the discussion will shift to who has the best Class D versus is Class D good. We believe we are leading the way on that. |
