Comparative images, same scene, with the 70-200VR alone and with all three Nikkor TCs: TC14E-II, TC17E-II and TC20E-II. Shot with a Nikon D2H on a Gitzo 1410 with Markins M10 ballhead using careful controlled technique. VR off, of course.

Shot as NEFS and rendered in Nikon Capture 4.4.2, with no additional processing, except the last 2 images as indicated. Sharpening is in camera "Normal".

The 1st four images are full frame, reduced for web viewing. The 2nd set of 4 images are 100% crops of the center area. Focus was on a chimney pipe in the center of all the frames. Just for fun I included a 2nd version of the 400mm shot with additional USM, but no In Camera sharpening, just to show that processing impacts these lens tests.

Shot with a D2H, so may not be representative of the higher res sensors most people are buying today (D200 D80). Also, since this is a fairly distant landscape scene, atmospherics come into play. It would be better to shoot a closer subject but this is all I have at the moment.

What you don't see is that in real world use, the consistency seems to decline as the TC is strengthened. This accounts for many of the real world variables that come into play with increasing focal length (and camera blur) and the availability of increasingly smaller apertures and consequently longer shutter speeds. I use the least powerful TC I think I can get away with and use the TC14E the most. If you are only buying one TC, I would get the TC14 unless you really think you need the extra reach.