The 35mm "Full-Frame" vs. DX size digital sensor controversy is caused by a general lack of "digital sensor" knowledge on the part of many users, and strong marketing propaganda on the part of some sellers.
Many camera users haven't investigated the design of lenses and how light must fall on a digital sensor, compared to film. Sound physical evidence shows that, due to the micro lenses over the sensor, a 35mm lens cannot fully cover current 35mm sized sensor chips! The light fall-off from 35mm sized lenses can be as much as two-stops on the frame edges, simply because a sensor chip requires light to hit it almost directly.
23.7 x 15.6mm JFET LBCAST DX sensor on the Nikon D2H and D2Hs
Look at it this way. If the sensor's pixels are like a deep bucket, they need to be hit directly from the top to enter. If you tossed a ball into a bucket, it will go in easily if it's dropped in from overhead. But, if you toss it in from an angle there is a good chance it will bounce off the rim and not go into the bucket. (Which is why the basketball leagues pay so much to those who can get the ball in the hoop regularly) Light behaves the same way as the ball. 35mm lenses are not designed to bring light directly into the bucket (pixel), but instead cause the light to come in at a greater angle progressively as you move toward the edge of the lens. With 35mm film there are no pixel buckets, just random silver grains, so angle doesn't matter as much.
In Figures 1 & 2 below, I have attempted to explain how this works. The bubbles on the front of the sensor are micro lenses that attempt to focus the incoming light rays on the best part of the pixel bucket. The pixel buckets are represented by the white rectangles in the blue sensor.
Notice in Figure 1 how the light rays (green) are coming from the back of the lens and are being focused correctly on the back of the pixel element by the micro lenses on the sensor. No light is bouncing off the rim of the pixel bucket.
As you move from the center to the edge of a 35mm lens a higher and higher level of light is unable to enter the pixel bucket directly, so the light drops off and you have less light on your subject on the edges of the frame. As we all know, the less light that reaches the sensor the harder the "chip" must work and the more noise that is generated. So, a full-frame chip with a 35mm lens will have more noise on the edges and lots of light fall off.
Notice in Figure 2 how the micro lens is struggling to focus the light on the back of the pixel element. Since it cannot cause the light to go in directly, there is a certain amount of light lost.
This is called light "falloff," and causes the edge of the image to be dimmer. The larger the sensor is, the worse the angle the light hits the edges. So a full-frame sensor is destined by the laws of physics to have significantly more light falloff than a DX sized-sensor. In many cases the falloff can be as much as two full stops of light.
There are three ways to lessen this problem:
Make the lens bigger so that it will have a larger image "circle" and the light will hit the sensor more directly.
Make the sensor smaller so that the existing image circle can more completely cover the sensor.
Redesign the lens so that it "collimates" the light, or makes the rays come out of the lens in a straighter path.
As more DX lenses are released by Nikon we will have even better images than before. But, in the meantime, all of your legendary and pro 35mm Nikkors can continue making their excellent images.
More articles that might interest you