rob watts clearly states the scaler and tt2 models were developed in the same time frame, in some ways the fpga code for conversion and filtering written into the tt2 is optimized for the scaler fronting it... otoh dave has even more computing power but was a predecessor in deveopment time frame by a couple years, not to say there isn’t benefit to be had with an upscaled input feeding it...
my experience is the simply that the scaler feeding the tt2 makes it smoother and more spacious sounding, yet with no loss of detail/micro-dynamics, transients are more effortless and cleaner/more natural - the term i have used in an earlier post iirc is that somehow the musical ’bits’ seem glued back together more like the real thing, more seamlessly, with the scaler working
conversely, removing the scaler makes the tt2 sound more hifi -- more treble energy but in a glare-y sense, music seems sharper and more uptight - on my system the difference is immediately noticeable, no need to strain to hear it (which can be the case in trying to hear the difference among other finer dacs, say for instance, weiss 501 vs bricasti m1se)