He may be misinterpreting it, but it would be better if it were noise. It may be worse than that. It was mentioned once off the bat by
@pma months ago and then promptly forgotten or ignored (although I gave up after page 35). This is nowhere close to state of the art--it's not even very good. This "normal distortion rise with frequency" (see comment on chart) was normal two or three decades ago. While many amplifiers still do that, the good ones
don't. Good amplifiers that use proper feedback schemes and/or don't run out of open loop gain so they can apply enough feedback do not. This rise is only "normal" only for old amplifier and badly designed newer ones. Follow the 15kHz line out to, say, 10W. The amplifier is creating distortion of .02%. Now go out to clipping, which we'll just call 80W. THD+N has risen to about
.08%. It's impossible to tell whether that is THD or N, since if there is distortion which is not harmonic, it would show up as N.
A simple CCIF IMD test would help reveal what is what, but that's a test Amir rarely does. The multitone helps, but because it is a multitone, it can also serve to mask distortions with another one of the tones. This is why Audio Precision counsels against (or at least, does not advocate for) using this as a distortion test or a test of an amplifier's linearity. The high frequency sweeps, though, at least give you *some* inkling that all is not rainbows and unicorns. There's a problem here, and this ain't no Benchmark or a Purifi. I have measurements sitting on my desk with a sweep for a 20+ year old lowly PA amplifier. At 20kHz (which would be a *slightly* worse case than 15kHz), THD+N is under .005%. A two decades old amp that can be had for a few hundred bucks is more linear that this chip amp by an order of magnitude. (To be fair, I bought this specifically because it does this and used a unique feedback topology that allows it to do this--it may well be the only PA amp capable of this.)
Taking some modern, available examples, the AHB2 at 5W is at .002% at 15kHz, rising to a worst-case of .003%. A NC500, .005% at 5W. That's Class D done right. A McIntosh MC462 will give you .0035% at 100W. See
https://www.stereophile.com/content/mcintosh-laboratory-mc462-power-amplifier-measurements. To be fair, the Behringer A500 measurements were at .13% at 15kHz, while this is at .02%, so at least it's better than that... Again, we don't really know how linear the amplifier is and how much trash it's *potentially* throwing into the audible spectrum because we don't have a CCIF 2-tone. But, amps that do well at 20kHz *ALSO* tend to do well on the 2-tone, and vice-versa. The inverse is also true. Bad 20kHz THD, bad 2-tone.
If we were ranking things in term of the amplifiers ability to stay linear at high frequencies, this is about on par with the lowly Onkyo M282 tested awhile back (graph cuts off at 10kHz, so it's not clear). But that will crank out 150W if you need it. It also costs about $300. Anyone that thing there's some sort of Purifi or Hypex grade performance here at a cut-rate price is just wrong. It's a one trick SINAD pony, whereas those better amplifiers appear to offer the whole package. Topping did a nice job of gaming the SINAD (THD+D) at 1kHz (as noted in the review itself with re: the gain structure). That was a nice trick. If they can manage to cure the inherent high frequency linearity issues of the Texas Instruments chip, it would be an amazing if not impossible trick. But they haven't done that and probably can't, so what they've done here just isn't terribly impressive at all, in my book. Ain't no such thing as a free lunch.
If I wanted to be sure I bought a highly linear and technologically superior amplifier without the potential for audible defects? I wouldn't buy this when its high frequency performance is mediocre, at best. In fact, that's why I
didn't buy this and bothered posting this. In short, the charts aren't showing noise rising at high frequencies. They are showing worse, and
probably showing that this amp does what these TI chips tend to do that Hypex or nCore do not: Get nonlinear and start spraying off distortion that is not harmonic at higher frequencies. Audible problem? Who knows, but I would prefer to have something that does not exhibit this problem. .08% THD at 15kHz at only 80W is just plain bad. Sorry for the book and the diatribe, but someone needed to set the record straight, and since I spent way too long reading this thread, I thought I might as well do something slightly productive with the investment of time..
Edit: Rate the amp using FTC specifications and it would be this: Less than .08% THD+N from 1/4W to 80W into 4 ohms, 20Hz to 20kHz. That's not exactly impressive. That's plain terrible. I have 40 year old amplifiers that blow that out of the water. The merits of the specification might be debatable, but it is what it is.