Hi Hugh,
Yes I think that was basically my point. Although independent of pixel size which as you point out is a big difference there is total amount of light the sensor receives simply by its size. Full frame is 2.25 times the surface area of APSc. Yes FF usually have larger pixels but some don't like Sony Nex 5 (16mp) and Sony A7r (36mp). The A7r noise levels are way lower.
It may be clearer this way - full frame image compared to tiny sensor in an Iphone image. The iphone has small pixels but also a tiny sensor. Noise levels are nowhere near the full frame image.
Micro 4/3rds cameras have excellent IQ until light levels drop and then the biggest complaint I read is noise even at low ISO levels. Yes small pixels but there is the factor of size of the sensor collecting the total amount of light.
There is a very clear Wikipedia article about this that explains it way better than I am. I'll try to find it again.
I agree the above gets lost because of the different sized pixels, different efficiencies of sensors and the general rule that large pixels have less noise and deeper wells than small pixels and thus lower noise.
Another way of expressing it would be APS sensors will have to work much harder and be more efficient to keep up with a full frame sensor.
By the way the Xtrans and Sony A7r are both Sony Exmor sensors so the differences there are not as great as you would think. So in a way they are a good example. A better example is this one:
http://www.dpreview.com/reviews/imag...62535884370765
The Xtrans colour filter array does complicate things but basically its designed to allow no moire with no antialiasing filter and to reduce luminance noise both of which it does and hence the marketing claim that their APS sensor can match full frame (not entirely true but it does do a good job).
Greg.