As I have understood by this question, an exposure value of a certain scenario cought from a digital sensor (CCD or CMOS) results from:
- the scenario luminance
- the sensor sensitivity
- the ISO, shutter speed and F number settings
Where ISO is not the sensor sensitivity (which is proportional to the sensor pixel area and to its quantum efficiency), it is just the tunable amplification factor.
When using an on board lightmeter, I guess it, once measured the (reflected) amount of light, it adds up the sensor sensitivity contribution to tell us if the image we are going to take will be properly exposed.
Well, this conversion factor is unknown to a dedicated external light meter. So, I guess its suggestions may be not optimal for all sensors, and may even depend on the utilized lens.
I'd say that a calibration procedure would definitevely solve this question. But why don't usually people even think at this? Is it a negligible effect? Would this effect cause less accuracy than the low accuracy provoked by the reflective measurement of a TTL on board light meter?
Will an old light meter, calibrated for old sensors with lower quantum efficiency, give us proper suggestions when utilizing an efficient camera?