There are two factors determined by physics that favor autofocusing with wider apertures.
- More light An f/2.8 lens lets in twice as much light as an f/4 lens. The more light an AF system has to work with the faster and more accurate it can be. Signal-to-noise ratio affects performance of the sensels on PDAF sensors the same way that it affects performance on imaging sensors. Sampling period for PDAF sensors affects SNR in much the same way that exposure time does for imaging sensors. With brighter light passing through a lens, the SNR will be higher.
- Wider baseline Phase detection AF works by comparing the differences between the light coming from the right and left sides of the lens for horizontally sensitive AF points, from the top and bottom of the lens for vertically sensitive AF points, and from upper right/lower left or upper left/lower right or both for diagonally sensitive AF points. The wider the effective aperture of the lens (more properly called the entrance pupil), the further apart the light rays that are compared can be.
To take advantage of a lens' wider aperture, the pairs of sensors for a particular focus point in the AF array must be further apart from one another. But that makes those lines not very useful when a lens with a narrower aperture is attached to the lens. So camera manufacturers hedge their bet a little. Some of the focus points are more sensitive/accurate but only function well with a large aperture lens. Other focus points are tuned to be able to use the light from lenses with narrower apertures. But those points can't take advantage of the wider light rays provided by a wide aperture lens. Some AF "points" have more than one set of lines that are sensitive at different apertures.
This is because the two lines on the focus array for each focus point are in a fixed position. If they are close enough to each other to be able to use light that gets through each side of the lens with a narrow f/8 aperture, they are not far enough apart from each other to sense the light that gets through the edge of the lens with a wide f/2.8 or wider aperture. Even when a faster lens is on the camera they are only using light falling on each side of the lens that is close enough to the center to make it through the narrower aperture.
How well the 70-200 f/4 takes advantage of your camera's focus system depends on the specific design parameters of your camera's focus system. In general though, a constant aperture f/4 telephoto lens will perform well. The only place you might be concerned is if you plan to use a teleconverter, since a 1.4X will raise the lens' maximum f-number to 5.6 and a 2X tele-converter will raise it to f/8.
Incidentally, the same physics that favors AF with lenses with wider apertures also favors DSLR cameras with larger sensors. Because the mirror is larger (particularly because it is wider) in a full frame camera than in an APS-C camera, the baseline used for the most sensitive focus points can also be wider.
For a little deeper answer on how cross type points work and a visualization of how f/2.8 points require lines that are further apart, see this answer.
From the comments:
Hi Michael, the answer explains about Canon and I have read quite a few documents over the internet on how cross type sensors work. I have noticed a pattern; while Canon puts more cross type sensors, Nikon engineers take a different argument for not putting that. They say if the light comes with certain angle, cross type sensors behave as a normal sensor. Anyways, what is your point of view on Nikon D610 with my 70-200 f/4 lens? Will the performance improve? I am also planning to use a teleconverter, but that's a distant plan as of now.
The physics is the same regardless of the manufacturer. All PDAF focus points use a pair of lines. Cross type focus points use two pairs of lines: one pair for the vertical and a separate pair for the horizontal. The reason you read more about this online in regard to Canon is that they are very open about which lines (horizontal vs. vertical) on which focus points are sensitive at what apertures for each of the focus systems of their cameras. At Nikon this is apparently top secret, at least for the D90. There's a reason most top sports photogs have been using Canon since the early 1990s: superior focus speed/accuracy.
If you are ever planning on using a teleconverter, especially a 2X, with a 70-200 lens you should consider an f/2.8 lens. With regard to the D610 there are two factors at work. The FF sensor means the FoV of the 70-200 will be wider than on an APS-C body such as your D90, where the lens acts like a 105-300mm lens in terms of FoV. You can always crop the result (or use Dx mode to let the camera do it for you), but you give up megapixels when you do that. The focus system of the D610 is much more advanced than the D90. 39 points, including 9 cross, with 7 sensitive @f/8... This means the f/8 "points" will work with more lenses, but also means they can't take advantage of the wider baseline provided by faster lenses.
One question, I could not understand your last comment "This means they will work with more lenses, but also means they can't take advantage of the wider baseline provided by faster lenses." Why so?
Because wider aperture lenses have entrance pupils that are wider. PDAF works by comparing the rays from the same spot that enter each side of the lens. Have you read the answer linked at the end of the main body of this answer above? Although it is in regard to the 7D the principle is the same: The wider the baseline the more sensitive and more accurate a focus point can be. And don't get caught up over cross type points. In reality they are just two points, one vertical and one horizontal, that cross each other. In the case of a diagonal cross point they are two diagonal focus points that are 90° apart.
Some have misquoted a post at DPR by noted optics expert Marianne Oelund to claim that all PDAF systems are limited to around f/7. Those misquoting her seem to claim that the following is true for all PDAF systems:
The aperture that controls how much light is received by the AF sensor is the aperture of the PDAF separator lenses which have been shown to be ~f/7. Having the main lens aperture larger does not increase the amount of light. The only time the main lens aperture is relevant is if it is smaller than the AF system is designed for (typically ~ f/5.6).
Nothing could be further from the truth (regarding the above statement applying to all PDAF systems)!
Many cameras have AF points sensitive all the way down to f/2.8 or wider. What is described in the quote above is the Nikon D300 camera introduced in 2007, a specific camera with sensitivity only down to f/5.6 of each of its AF "points". But not all cameras are like that. In fact, most current DSLRs are not. They almost all have some AF points that have a wider baseline than what is described above. From near the end of the fourth post in the DPR thread, also written by the quoted OP, Marianne Oelund: "In Nikon's AF systems, the separator-lens images are set just inside the f/5.6 circle." That was not even current for Nikon DSLRs as of 2014 when it was posted, but it had been mostly correct until shortly before 2014. Marianne's statement here must be taken in the context of her introductory remarks in the original post to the DPR thread:
This thread will present the stepwise development of a phase-detect
autofocus system...
and
"See the following posts for presentations of each step in the
development. More posts will be added later, as I have time, and/or in
response to questions."
At one time Nikon tended to make all of the AF points in their cameras functional at f/5.6 or f/8 and those systems did not give any increased sensitivity with larger apertures. Canon, on the other hand, has been including AF points with various sensitivities for far longer. Nikon does that now as well, although still not to the extent found in many Canon cameras.
From another comment (italics added) by Marianne Oelund later in the same DPR thread:
The AF doesn't "operate" at f/5.6 DOF. It uses the f/5.6 circle of the
main lens exit pupil as a baseline to triangulate from, but yes, this
isn't a very wide baseline and that's why some camera companies offer
f/4 or f/2.8 AF points.
If you go way into the thread to Marianne's post titled "EOS-1Dx AF Sensor and Optics" you can clearly see the difference between the Nikon D300 (a 2007 model) and the Canon 1D X (2012). But Canon was doing it way before 2007.
Yes, focus points designed for f/2.8 have greater accuracy. But they only operate/function with lenses of f/2.8 or wider which is not necessarily a benefit. I.e. in earlier Canons they had an AF sensor that had vertical sensors effective at f/2.8 and horizontal sensors effective at f/5.6 (i.e. 1D M IV)... the result is that with lenses slower than f/2.8 the vertical sensor lines are masked off and the focus points can not function as cross types.
The benefit of having all of the sensor lines w/in the f/5.6 objective area is that cross type performance is retained for most lenses. The negative is that the angular offset is less and therefore the points are less sensitive (the virtual images move less). A focus point that is designed for a smaller aperture (i.e. f/8) remains effective at larger apertures (i.e. "all the way to f/2.8 or wider"), the opposite is not true. Still, my comment was only in regards to the amount of light at the AF sensor chip as transmitted by the separator lenses.
The original (mis)quote above implies that all cameras cut off at between f/5.6 and f/8. That simply is not true. Neither is the vast oversimplification of Canon's approach. With the early AF systems some of the 'points' were vertical and some horizontal. The early cross type points were typically only in the center and were sensitive at f/5.6 in one direction and f/4 in the other. When Canon introduced f/2.8 dual cross type points they were a diagonal f/2.8 sensitive cross-type pair of lines superimposed on an f/4/-f/5.6 pair of lines in the vertical and horizontal orientation.
The earliest f/2.8 AF "points" from Cannon were, in fact, four sets of lines - one horizontal line sensitive to vertical contrast at f/5.6 or wider, one vertical line sensitive to horizontal contrast at f/4 or wider, and two crossed diagonal lines sensitive only at f/2.8 or wider. The performance of such AF systems is one reason Canon dominates the sports/action segment.
But they only operate/function with lenses of f/2.8 or wider which is
not necessarily a benefit.
It certainly is a benefit if you are using an f/2.8 or faster lens!
An f/8 AF "point" may remain functional when used with an f/2.8 lens, but it will by no means be as effective as an f/2.8 AF "point" when used with an f/2.8 or wider lens.
A bit more background to hopefully clear up some misconceptions still floating around this community.
The simple fact is that the PDAF module uses small images taken from small areas of the objective lens which creates a "virtual aperture" as discussed in the DPR article. It simply cannot be equivalent to the lens' max aperture.
No one is saying any part of the AF module is getting all of the light passing through the lens. Even f/8 AF "points" don't get all of the light passing through an f/8 lens! However, the location of those small areas that are sampled is wider. That is, the locations sampled are aimed further apart on opposite sides of the lens for a set of microlenses and AF sensor lines used by an f/2.8 AF "point" than the locations of the small areas for a set of microlenses and AF sensor lines used by an f/4, f/5.6, or f/8 AF "point", even on the same PDAF sensor.
AFAIK, all PDAF systems are designed primarily around an f/5.6 lens
aperture so that there will be no shading/vignetting/light loss at
f/5.6 or wider... i.e. f/5.6 is the maximum light potential to the
module.
This is not correct. The specific AF system of the Nikon D300 referenced in the initial information in Marianne Oelund's DPR post was one such system that does match the description above. But as we have already demonstrated above, Marianne herself acknowledged further on in the discussion that not all PDAF systems are equally narrow or equally uniform from one AF point to the next.
If the statement above were correct, then f/2.8 AF "points" would still function to some degree with an f/5.6 lens. The fact of the matter is, such f/2.8 "points" do not function at all with f/5.6 lenses. Why? Because when an f/5.6 lens is used with a PDAF system that incorporates f/2.8 sensitive lines, the f/2.8 microlenses are pointed to a spot outside the edge of the light the f/5.6 lens allows to pass.
This is why AF "points" with diagonal cross type f/2.8 lines also include another set of horizontal/vertical cross type lines for the same AF "point". (A cross-type "point" is really one horizontal and one vertical set of sensor lines that both point to the same spot in the lens' FoV, such that the areas of sensitivity for both lines is superimposed on the same area and form a cross. When a diagonal dual cross-type "point" is added, that makes a total of four sets of independent lines all aimed at the same spot in the camera's FOV.)
Here's a graphic that shows which parts of each AF "point" are sensitive at which apertures for the EOS 1D X/5D Mark III PDAF system:
There are multiple sets of "splitter" microlenses on PDAF systems. Here is the microlens array that sits over the AF sensor for the simple "9 point" AF system of the original Canon 5D:
Compare that to the much simpler splitter array on the front of the Nikon D300 AF sensor (the camera referenced in Marianne Oelund's DPR post):
It should be fairly plain that not all AF sensors are made the same way with the same characteristics.
Although PDAF systems with higher numbers of "points" often share portions of longer sets of sensor lines and the microlenses that project light onto them between multiple "points", the more complex 19, 45, 61, or 65 "point" systems still use more complex sets of microlenses than the array for a 9-point system that is pictured above. Each set of microlenses and sensor lines can be more restrictive, the same, or less restrictive than other sets of microlenses and sensor lines used by the same PDAF array. It all depends upon how wide each set of microlenses are pointed (towards the lens) and how long and how far apart each set of sensor lines on the PDAF sensor proper are from each other under the microlenses.
The 65-point AF sensor for the Canon EOS 7D Mark II. Notice how much further apart the sets of diagonal lines are. They are for the diagonal portions of the f/2.8 sensitive diagonal dual cross-type center AF point.
As indicated by text in the above image, included in a press release from Canon upon the announcement of the EOS 7D Mark II, the limit of the total camera/lens system is determined by the maximum aperture of the lens when it is narrower than the PDAF array's baseline for specific AF points.
The 61-point AF sensor for the Canon EOS 1D X and Canon EOS 5D Mark III. The sensor for the EOS 1D X Mark II and EOS 5D Mark IV are very similar.
Notice the ten sets of diagonal sensor lines spread further apart than other sets of lines on the 1D X/5D III AF sensor. They are for the f/2.8 sensitive diagonal elements of the five dual cross-type "points" in a vertical line at the center of the AF sensor. These AF "points" also include independent f/5.6 sensitive horizontal and vertical lines for the same AF "point".
Here is a graphic that shows the various AF "points" for the Canon EOS 7D Mark II (65-point) and Canon EOS 1D X/EOS 5D Mark III (61-point) PDAF sensors.
Some have additional points that can use the f/8 or f/2.8
objective areas, but that has no effect on the amount of light/size of
the virtual apertures.
But it does have an effect on the difference in distance between the two samples taken from each side of the lens. Microlenses and sensor lines for f/2.8 AF "points" are aimed wider than those for f/4, f/5.6, or f/8 lenses. This allows them to be more sensitive, as the difference from one side of the lens to the other will be greater for the same defocus distance than for points that sample two areas closer to each other and the center of the lens.
