I am curious why the measurements for Filter 4, Figures 5 and 15 (Digital Edition), show a higher noise floor than the measurements for the other filters? Does Filter 4 really cause the noise to be higher than the other filters or am I misinterpreting something? Compare to Figures 4 and 14.
A higher noisefloor with high-level, high-frequency tones? Yes, the noisefloor is much higher than with the other filters. As I explain in the text, this filter is "leaky," ie, offers very little suppression of ultrasonic sampling images. So what you are seeing is the effect of the leakage of image energy, as ultrasonic products fold back into the audioband. What I find surprising is that some people find this kind of behavior preferable to that of a better-behaved reconstruction filter - see my comments at www.stereophile.com/content/2011-richard-c-heyser-memorial-lecture-where-did-negative-frequencies-go-case-study-3-digita.