Nice catch.Digital Loop Filter being AX5689.
Amirm, in your review you say "Either way the input to the amplifier is digital "
??? I don't see a digital audio input, though. Is there one?
We already know. This is worse, appart from THD+N Vs Frequency.
Because they do need to sell M400 and SU-9 DACs.
AudioXpress has an 2017 article on the AX5689. It has a few performance charts from Axign with its chip paired with ST516B, TAS5342A and TAS5624A, but for lower power output (~50 W @ 4Ω). The Axign THD+N numbers looked to be about 10 dB better those Amir measured, and seemed to be noise related. (Again, the Axign implementations were lower power, so not exactly apple-to-apple.)
Thanks for this article. Taking in account this information I can understand why they choose the ST516BE but still we see that the chip is having a hard time controlling the noise of the output stage, may be a dual ST516BE would have done the trick (the SMLS dual MA12070 is way much better than the single chip version) or an infineon output chip like the IRS20965S or the one used here https://www.digikey.se/catalog/en/partgroup/irs20957-evaluation-board/1599Interestingly, according to the article, the AX5689 + TAS5342A performed better than the AX5689 + TAS5624A, when the AX5689 didn't have the "wrestle with the internal feedback loop of the TAS5624A". So, it may actually perform worse with the TAS5612A that has internal feedback. (The performances of ST516B and TAS5342A were identical in the article.)