Let's continue the theme of the original post and review some other measurements atomicbob has made. In this case, Schiit Modi 2 Multibit. He prefaces the review with this text:
Here is his test of -70 dB sine wave:
Focus on the green curve. That should be a perfect looking sine wave at this level yet it has chewed up sections. Things get a lot more interesting at -90 dB:
The green curve has no resemblance to anything called a sine wave. Here is Topping DX7/DX7ss:
We see a very close match to a sine wave compared to total garbage that Modi 2 Multibit produced according to Bob's measurements. The -90 dB sine wave in Topping DX7/DX7s is cleaner than the Modi 2 multibit at -70 db!
With -90 dB within the envelope of a 16 bit signal, it is not fathomable how he can say this is "incredible performance for 16 bit DAC." The performance does not at all reach 16 bits. We can see this in his linearity measurements:
He is using a whopping 30 dB scale (+20, -10) which helps to flatten the significant error here. See how the error starts very early as low as -70 dB. That means it can't produce 12 bits of resolution correctly.
The error is more dramatic when we look at S/PDIF input as opposed to USB above:
This is what we get out of Topping DX7/DX7s:
There is just no comparison. One design is broken and the other is not. Yet we have commentary from Bob that paints an entirely different and incorrect picture. No wonder people think these are nice DACs despite the data. The data is obfuscated by words that don't match the measurements.