7D Speed of AF Servo "lock"

While reading another thread (https://photography-on-the.net/forum/showthre​ad.php?t=904086) I understood it to say that it takes between 0.5 and 1 sec for AF Servo to lock.

Is this correct?
Does this time vary dependant on what lens is being used?
Does it depend on lighting conditions?
What factors affect the length of time it take to lock?
What, if anything, can be done to reduce up this "locking time"

Does this time vary dependent on what lens is being used?
--- YES , my 35-350 and 50 1.8 are the slowest lenses I have [focus-wise] 17-55 is the fastest
Does it depend on lighting conditions?
--- YES - not enough light and it starts hunting
What factors affect the length of time it take to lock?
--- lens , light , camera , hot many focus points you pick [Spot - evaluative etc] , how much change in color at focus point
What, if anything, can be done to reduce up this "locking time"
--- The camera is fine , a newer lens - [not new] with faster focus , spot focusing instead of evaluative , The 7D has a larger focusing area in spot focus so use space between eyes instead of eye , learn camera .
Play with camera . maybe even write down what you do .
You don't expect to get in a new car and just set time on radio -

What you are referring to is the time it takes for the camera to establish a statistical database of readings from a moving target. This is then used to predict the position of such a target when a picture is taken, something which occurs some time after pressing the trigger button. The camera will then use its predictive algorithm to calculate where to pre-set focus on the lens, in order to make the subject come into the focus plane when the shutter opens.

For the AI Servo AF II algorithms, implemented first in the 7D and now also in the 1D Mark IV, Canon first claimed that there was no longer any need to pre-track a subject for this prediction to work. But in the most recent documents, they seem to have reverted to the opinion that giving the camera half a second or so of tracking boosts performance.

Apart from that, cameras like the 7D and 40D, provided with a lens with a fast focusing mechanism, like the EF-S 17-55 mm f/2.8 IS USM, are about the fastest their is, when it comes to measuring a focusing condition and then fix it.

Thanks for the replies guys.

I mainly photograph sport, high school soccer and adult rugby. I use a 70-200 F4 IS on a 7D.

My results have been inconsistent.

I have blamed the "learning" curve for the missed focus but based on what has been said above I am probably expecting too much out of the AF system. I never realised that 0.5 to 1 sec was required to achieve lock.

I will try predict the game more to allow for this "lag" required to lock focus.

This fact needs to be pointed out to people who complain about the 7D's AF system.

As an after thought, how long does it take achieve focus in one shot mode?

One shot is pretty damn quick

Overall i must say i must either be very lucky or very understanding of my 7D's AF or something...

My fastest lens is my 15-85 I think, Followed by a close race between the 100-400 and the 100 f/2.8L (with focus limiter on)

My old 1DII seems to require little to no time to achieve a lot. My 40D requires some and my Rebel T1i really needs a noticable amount of time to lock. If you don't get lock and take a sequence of shots then it will often never catch up, it will just lag behind the whole time.

davethejnz wrote in post #10560519
I have blamed the "learning" curve for the missed focus but based on what has been said above I am probably expecting too much out of the AF system. I never realised that 0.5 to 1 sec was required to achieve lock.

Did you read what I wrote above? The time of about half a second or so is to allow the camera to collect enough samples to be able to accurately predict the speed and direction of a moving target. Since the camera is capable of tracking a racing car at surprisingly short distances, targets like someone running across a field, with the magnification achievable with a 200 mm lens, doesn't take too much of this predictive capability. Your problems are probably more related to the question of properly selecting the AF point configuration and other AF-related settings (the first few functions in Custom function group III) than this particular feature.

It all depends. You have to define light level, distance to target, distance set on lens when focusing begins, lens used, reflections in the lens etc. to be able to give an appropriate answer.
I've seen a test where the lens was set at infinity, the camera aimed at a target 2 meters away and the focusing time measured. This was with the EF-S 17-55 mm f/2.8 IS USM. Cameras like the 40D, 50D and 7D all got the same time, 170 ms. Which shows that for this task, they all have enough calculation power, so it's mainly lens drive time that's the time consuming task. This test is done by Swedish magazine FOTO for each camera they test. The time, 0.17 s, has not been beaten by anything, not even the 1D Mark IV, but that did of course have to use anoter lens than the 17-55 mm. Comparable Nikon cameras usually need about twice the time.
To achieve initial focus lock in Servo AF is equivalent, or actually a fragment faster, since the exposure time used for the AF sensors is shorter in Servo AF than in One Shot AF. But once again, this is something different than the sequence of measurements the camera needs to build the database needed for predictive Servo AF to work.

That .17s is lens drive time from infinity to 2m. That doesn't have anything to do with AI SERVO lock for tracking. An intelligent photographer with a good lens will preset it to the distance they are expecting to shoot at anyways so the lens doesn't have to drive all the way in or out. This doesn't always work of course, you may be surprised by something closer or farther than anticipated but the test for AI SERVO lock would be much better if you took the lens drive out of the equation entirely, if you started the test focused on an object which then suddenly starts to move toward you.

This is a common shooting situation but is very different from airplanes and birds already in flight, racecars already in motion, etc. A perched bird which suddenly takes off, a sprinter who is just leaving the blocks, etc. are a very different situation and it is here, where the subject suddenly begins new motion or changes direction that my old 1D seems to be much faster than my 40D and Rebel T1i. For things already in motion the 40D and T1i do fine, a bird already in flight is usually quite predictable unless they are catching insects in mid air in which case the twists and turns will fool AI SERVO. In field sports with players starting, stopping, turning, changing direction it is also much more challenging as the computer has to throw out its existing model and build a new one.

hpulley wrote in post #10562757
That .17s is lens drive time from infinity to 2m. That doesn't have anything to do with AI SERVO lock for tracking.

No, but there was a question about One Shot AF as well.
I've not used any 1D camera for anything but a few test shots, so I can't compare there. On paper, reading the specifications, the 7D is the only EOS camera outside of the 1D range that has the same, or at least similar, capabilities as far as AF goes. But again, I can't compare.