There’s been much speculation about Sony introducing a NEX camera with a 36mm x 24mm “fullframe” sensor. This has been accompanied by angst and naysaying, much of it technically unfounded. One of the objections to the possibility is the perceived geometric impossibility to cover the whole frame since it looks like the mount is too narrow. Even though the NEX-VG900 camcorder has a 36mm x 24mm sensor and E mount, and people have successfully reported to have use all sorts of “fullframe” lenses on it, some people still seem to suffer from cognitive dissonance when looking at a photo of the sensor/mount as shown on the left. People seem to by hung up on the fact that the corners seem to be occluded.
The concern may stem from a mistaken belief about the path that light takes to get to the sensor. One must realise that the light passes through the lens and ultimately through the rear element before being projected onto the sensor. The light which exits the rear element will diverge to reach the extremities of the sensor. The rear element is usually smaller than the throat, even on a large aperture optics like the Minolta 100mm/2 or 85mm/1.4 G the rear element is only ~34mm in diameter. The throat of the E mount is 46.1mm in diameter, the fully clear diameter is 43.5mm, the maximum width of the rear element is 41.5mm (taking into account the barrel housing), and the fully circular region (excluding the contacts) is 35.5mm.
One may object that the 18mm flange distance of E mount is much greater than the 44.5mm of A mount, hence comparisons are moot. One just needs to look at the Leica M mount to see that it should not pose too much of a problem. The M mount has a diameter nominal diameter of 44mm (c.f. 46.1mm) and a flange distance of 27.8mm, considerably smaller than SLR mounts. Leica Thread Mount (LTM) has a nominal diameter of 39mm.
There are issues which may complicate matters but they do not prevent the E mount from using a “fullframe” sensor. The most serious one is making sure that the light rays, as they leave the rear element, are not at too great an angle compared to the normal of the sensor surface. This is because the pixels have an angular dependence on the sensitivity. This means that it is optimal to have a telecentric lens where exit pupil is as far from the sensor plane as practical. A fully telecentric lens would have the exit pupil at infinity and hence the rear element would have to be at least as large as the sensor, but it is more practical to accept some small deviations. We can calculate the minimum angle of incidence onto the sensor given the dimensions of the E mount. If we use the fully clear circle of 35.5mm, the diagonal of 43.3mm for a “fullframe” sensor, and the flange distance of 18mm – 4.7mm = 13.3mm (taking into account the projection of the lens into the body, we have a minimum angle of incidence at the corner of 16.3 degrees. It should be easy enough to design the pixel to accommodate this and even greater angles. Leica use offset microlenses at the corners to compensate for this effect.
In summary, there is nothing preventing Sony releasing an E mount NEX camera with a “fullframe” sensor along with accompanying lenses. The NEX-VG900 demonstrates that there is not only possible, but has been achieved for all practical purposes. To get good corner performance, care must be taken to achieve adequate telecentricity, but this is true of all digital sensors. The pixel architecture and microlens layout will have to be optimised to ensure good angular response to off-axis rays. Lens design may need to compromise on compactness to achieve the requisite telecentricity. Adapted lenses on the VG900 have shown that there are no corner shading problems. Now all we have to do is wait and see if all the rumours are correct.
PS David Kilpatrick at Photclubalpha posted about the possibility of E mount “fullframe” some time ago.