Actual 40D speed test. And the results are...

In total, I ran three series of camera speed tests for the 40D shooting in high speed mode. Here are the camera settings I used:

Series 1 & 2: Av with f=4.0 so shutter speeds ended up being 1/1600”, ISO 1600, High Speed (6.5 fps), picture quality of fine large jpeg only (no RAW), PNY Optima Pro x133 GB CF card. I had all noise management turned off.

Series 3 was the same as above except I was in M to force a shutter speed of 1/1600” and an aperture of f4.

Here's the data for all three series, albeit only Series 3 seems meaningful - keep reading for the reason.

The first column, Time (sec), is the time in decimel seconds of the stopwatch in each photo. The second column is the time delta, in seconds, between two consecutive photos and the third column, is the calculated Frames Per Second (fps) based on the two consecutive photos and one time interval.

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

Here's a sample photo:

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

Well, I nominate you for Measurebater of the Month.

Mark_Cohran wrote in post #4732091
Well, I nominate you for Measurebater of the Month.

I'll second that.

Mark_Cohran wrote in post #4732091
Well, I nominate you for Measurebater of the Month.

Thank you, thank you, but stand back and hold your applause 'cause I'm usually good for at least two.

Phil's review on dpreview shows a plot of frames/second achieved as a function of shutter speed. Not surprisingly, the frame rate slows down as your shutter speed gets longer. At a shutter speed of 1/15, the frame rate has dropped to 4 frames/second.

Steve

A simple test is to simply hold the shutter button down until the buffer is full then press the info button and check the time stamp on each photo. Count the number of frames shot in a 2 second period, divide by 2, and you should have the frames per second.

Sparky98 wrote in post #4733800
A simple test is to simply hold the shutter button down until the buffer is full then press the info button and check the time stamp on each photo. Count the number of frames shot in a 2 second period, divide by 2, and you should have the frames per second.

Sorry but this doesn't work. In my tests the third, 14 frame test, only took 0.01 second according to the time and date stamp on the photos. Basically the camera quits updating the clock when it "burst" mode.

My camera clock doesn't quit though it only records to the second and not in tenths or hundredths. I managed 21 shots until my buffer filled up. I had either 12 or 13 shots in a 2 second span, it just depends on which second you count the last shot. That is not exact but pretty close to 6.5 fps.

I had noticed this once before when I took some shots of a flag in a high wind. I took about 10 shots just playing around with high speed and when I reviewed the shots I noticed I had 6 shots all on the same second.

Did you trigger your timer electronically or by hand? Unless you have a device to trigger the timer and camera at the same time then you are introducing an error equal to your reaction time in starting and stopping the timer. I guess it sounds like I'm trying to put down your test method but that is not what I am trying to do. I am sure your method is much more accurate than mine but mine is a kind of quick and dirty way to see that a camera is operating close to the factory specs when one doesn't have access to more accurate test methods.

Sparky98,
The EXIF is intact on the photos I used in the test. The Canon 40D's clock records down to the 1/100 of a second but it was not updated during the high speed shooting.

The timer doesn't need to be synced with the start of the test because I am only using the change in time between the individual shots. I know my 14 shot test took exactly 2.00 seconds because the test started at 7.09 seconds and ended at 9.09 seconds. There is no error introduced.

I didn't use the EXIF data I just view the LCD on the camera and there was a time stamp on each shot. Six of the shots were on the same second and 6 or 7 on another second. What I was doing may not have been a valid test at all it just seemed simple.

Does having it in M mode make a difference though? The camera must require some time between shots in Av or other metering modes to calculate exposure. If this time is significant, and it uses the same processor for this as it does to process the image just shot, then it could affect burst mode times. So the discrepancy might be real rather than a problem with your GPS stopwatch.

You know what computers are like for calculations! The exposure is probably calculated in about a millionth of a second. One important factor in the reported test is using the lens wide open, as stopping down the aperture would slow things down a little. Also, I seem to remember reading that Canon themselves say the very fastest frame rate requires shutter speeds of 1/4000 or faster.

But if you're using something like evaluative metering then I would guess it requires some fairly meaty processing from the DIGIC... it probably is very quick but then in burst mode it might be queued behind the processing of the previous shot.