Mine:
- with 21/4 skopar, it underexpose around 1ev.
- with 35/2.5 skopar, UE around half~1ev,
- with 35/1.4 nokton, mostly correct.
- with 50/1.5 sonnar, mostly correct.
I tried to see the pattern of metering by pointing the camera to a light source, and see when it change the metering value. The result is showing me that it is really center weigted (eventough the sensor is on the side of "Mirror box")
I believe the UE is because of acuracy of stopdown metering on "all" RF is decreased when you choose smaller aperture.
In dslr/slr (mostly wide open metering), aperture is kept open wide eventhough you select smaller aperture value. The camera still measure fully open aperture, and compensate the result of metering based on aperture you choose.
compared to stopdown metering, when you select smaller aperture, the light entering metering sensor decreased (in slr, it doesn't) because aperture stops down. When it reach the sensitivity limit of the metering sensor, you will not get correct exposure. This even worse when the light is dim/low. The stopdown lens still cut the light when you stop it down.
That's why I can say, metering accuracy is going to be better if we have/select bigger aperture, or at least you have enough light to stopdown.
I believe, the angle of which light enters the ccd sensor is playing a big role as well, that's why, my 21 give me UE, compare to 35nokton, on the same light & aperture selection.
hope this helps.