i translated that review the best i could, cant say i agree with the comments
"The distortion values are outstanding...The F6 takes measurements at full level. Total distortion of about 0.0083%"
"The distortion spectrum shows that despite highly developed 32-bit technology
there is a certain difference in quality compared to expensive studio converters.
However, the performance is very decent for a mobile device."
assuming that number is THD+N, thats not all that great, it is a SINAD of about -81dB, which is more like ultraportable/handheld territory and step below what a balanced input recorder should be
it would be cool if you can link the rest of the pdf so we can see their methodology
the graph in the linked PDF shows the same low-frequency distortion that amir measured
when i try to measure the F6 anywhere near its stated max line input level of +24dBU the distortion blows it out. with a 1kHz sine wave, i can get up close to the max input (well to +22dBU which is the limit of my DACs output...i could throw a preamp in there to push it to +24 with equally disastrous results)
But cant get anywhere near that with the low frequencies
heres what the RMAA test file looks like. the two highlighted portions of the waveform are the low frequency portions of the sweep and multitone portion
note that the undistorted level of the original file has all the frequencies at the same level in the sweep and multi-tone portions
here are recorded 32/96 wav files at different input levels. from top to bottom, i tested input levels from +16dBu to+22dBu in 1dB increments. i didnt bother dialing in the gain on the zoom as it really wouldnt matter, so these are slightly off in overall dB but look at those low frequency potions relative to the rest of the waveform.
+16 and +17 dB look good, both look like this
at +18dBu we are starting to see distortion of the low frequencies of the multitone
at +19dBu we are seeing these get worse. the zoom is flashing the 'exceeding input level' on the screen
zoomed in to distorted waveform
at +21dBu the low frequency portion of the sweep is starting to distort as well
this gets even worse at +22dBU
to eliminate any variables of playback system, here is same playback system fed to the AD2K (now 20-year old technology) at +22dBu input. look at the absence of visible low-frequency distortion and how much cleaner the waveform is overall