Canon 5D spot metering area: 3,5% or 1,4%?

[_GUI_]

I have just purchased a 5D and was doing some tests to find out the effective circle radius accounting for spot mettering. I was quite surprised when reading in the user manual that the 5D considers 3,5% of the viewfinder surface for spot mettering (IMO 3,5% can hardly be considered a 'spot' anyway), but was more surprised to see this does not correspond with the measures I did.

I plotted a white circle in the center of a full black screen, and started to zoom out to find out at which circle size the spot mettering started to decay because of the influence of the black borders entering the spot mettering area.

The spot metering remained constant for any radius of the white circle larger than the smaller circle in the following image. That circle represents just a 1,4% of the whole sensor area. If that is correct that would be good news.

The second circle represents the exact size and position of the viewfinder's circle, which is claimed by Canon to be the reference for spot metering, but that circle is 2,4% of the sensor area, still quite smaller than 3,5%. Funily, this circle is not perfectly centred with respect to the sensor, specially in the vertical direction (see sensor axis), so I wonder if the spot mettering is centred with respect to the sensor, to the viewfinder's printed references, or to none.




And this is the sequence of metered light once we go to smaller circles than the apparent effective spot metering area:



Can anyone give any explanation about the figures obtained for the 5D spot mettering?

Regards.

PS: I did some tests over clouds that lead me to think that the effective circle considered for spot mettering is in fact smaller than the viewfinder's circle. After spot mettering over these clouds, there was nearly 3,5EV of headroom left from highlights to saturation:






Aprox. the same amount of headroom I obtained when spot mettering over a uniform surface with a tele lens and only considering the central patch, which should correspond to the theoretical maximum overexposure allowed from the camera metering to saturation:



If the effective circle when spot metering the clouds would have included those slightly darker clouds surrounding the brightest central area, there should have been less headroom as I had in other scenes where the highlights were not so well defined.

BR

[Colin Southern]

This circle is not perfectly centred with respect to the sensor, specially in the vertical direction (see sensor axis), so I wonder if the spot mettering is centred with respect to the sensor, to the viewfinder's printed references, or to none.

My guess would be "to neither". Obviously it's the intention to have it "under the spot", but the spot engraving is none-the-less somewhat "mechanically disassociated" from the metering, and also the sensor doesn't enter into the equasion at this point either. In other words, probably just a bit of mechanical misalignment creeping in.

If you check your AF point accuracy you'll almost certainly find a bit of variation there too.